Deep Fakes: AI for Good or AI for Evil?

1
00:00:01,700 --> 00:00:03,500
Hello.
Bizarre story, friends, you've

2
00:00:03,500 --> 00:00:06,800
tuned in to the bizarre AF where
we talked about the strange, the

3
00:00:06,800 --> 00:00:09,900
Absurd, the unknown and things
that are Pasar ass.

4
00:00:09,900 --> 00:00:12,000
Fuck.
I'm Kevin I'm going to be your

5
00:00:12,000 --> 00:00:15,000
host for today's episode and as
always we ask that you keep an

6
00:00:15,000 --> 00:00:17,200
open mind.
Keep us skeptically are but keep

7
00:00:17,200 --> 00:00:19,700
listening to those facts.
As we take you on our newest

8
00:00:19,700 --> 00:00:25,700
Journey, deep fakes AI for good,
or AI for Eve.

9
00:00:25,800 --> 00:01:04,000
Mmm,
Well, hello.

10
00:01:04,000 --> 00:01:07,600
My leash little darling.
How are you doing today, girl?

11
00:01:07,600 --> 00:01:11,000
I'm doing so good looking so
futuristic.

12
00:01:11,800 --> 00:01:14,200
I'm a Trekkie.
Yes, nice.

13
00:01:14,200 --> 00:01:17,300
Are you have you beamed down
from the Enterprise?

14
00:01:17,500 --> 00:01:21,700
I would hope.
So, yeah, I'm just here on like

15
00:01:21,700 --> 00:01:24,500
an away Mission away - yeah.
Good.

16
00:01:24,500 --> 00:01:25,900
Yes.
Thank you for joining me on your

17
00:01:25,900 --> 00:01:27,800
way mission today.
Thank you.

18
00:01:27,800 --> 00:01:30,400
You look great.
It's kind of like, Tron a score.

19
00:01:30,400 --> 00:01:35,000
Like like what's The what's the
Indiana?

20
00:01:35,000 --> 00:01:36,500
Jones Guy?
What's his name?

21
00:01:36,600 --> 00:01:39,100
I don't know.
What's the futuristic movie that

22
00:01:39,100 --> 00:01:40,700
he did?
Come on people.

23
00:01:40,800 --> 00:01:45,000
Oh running man.
No Blade, Runner Blade Runner.

24
00:01:45,000 --> 00:01:46,900
That's it.
Yeah, Blade Runner.

25
00:01:46,900 --> 00:01:49,100
Yeah, yes.
I like the crystal look like

26
00:01:49,100 --> 00:01:52,200
Priss.
Did you ever see that movie like

27
00:01:52,200 --> 00:01:54,200
years ago?
I don't remember any of it for.

28
00:01:54,200 --> 00:01:56,000
This is the one that had the
black stripe like that.

29
00:01:56,000 --> 00:01:58,200
And she was all like with the
white hair.

30
00:01:58,300 --> 00:02:01,400
She's all like you're giving me
like the makeups kind of giving

31
00:02:01,400 --> 00:02:05,400
me like dystopian.
And, which is, you know, yeah,

32
00:02:05,400 --> 00:02:07,900
that's the thing.
What is, what is that movie with

33
00:02:07,900 --> 00:02:11,200
Charlize Theron and oh my God,
Max.

34
00:02:11,200 --> 00:02:13,900
Yes, yeah.
Kind of love her so much.

35
00:02:13,900 --> 00:02:16,000
Oh good.
Well, thanks for joining here.

36
00:02:16,000 --> 00:02:19,600
Cheers.
So, cheers to a dystopian

37
00:02:19,600 --> 00:02:21,100
future.
Mmm, sure.

38
00:02:21,200 --> 00:02:27,000
We're not, hopefully not, I
believe in, I believe in Star

39
00:02:27,000 --> 00:02:32,600
Trek and ya know, to a start
tracking, Computer-assisted

40
00:02:32,600 --> 00:02:34,800
Future.
Okay, Okay.

41
00:02:35,100 --> 00:02:39,000
Humanistic like humanism like we
believe the best in people.

42
00:02:39,000 --> 00:02:40,600
Yeah, yeah.
Yeah.

43
00:02:40,600 --> 00:02:43,000
Whole Star Trek.
Yeah, that's not go any further

44
00:02:43,000 --> 00:02:44,300
than that.
Okay.

45
00:02:44,300 --> 00:02:47,200
Cuz cuz then it can get scary
images are going to talk about

46
00:02:47,200 --> 00:02:53,700
today.
So, Alicia have you Have you

47
00:02:53,708 --> 00:02:55,400
ever heard of deep?
Fakes.

48
00:02:56,400 --> 00:03:00,500
I have I've heard of deep fakes.
I thought that they were weren't

49
00:03:00,500 --> 00:03:04,200
they being used like to pretend
like you know presidents were

50
00:03:04,200 --> 00:03:06,800
saying things like Obama was
saying things that he never

51
00:03:06,800 --> 00:03:09,500
said.
And like yeah I recall like

52
00:03:09,500 --> 00:03:11,800
recall all of that.
Yeah.

53
00:03:11,800 --> 00:03:14,500
Yep.
So you have you ever seen them?

54
00:03:14,500 --> 00:03:15,500
Yes.
Yeah.

55
00:03:15,500 --> 00:03:20,800
I've also seen them with Tom
Cruise, like conkers doing like

56
00:03:20,800 --> 00:03:22,400
those magic tricks.
He's stuck.

57
00:03:22,400 --> 00:03:24,300
Yeah.
We're gonna talk about that

58
00:03:24,300 --> 00:03:25,900
little bit.
Yeah, but yeah, exactly.

59
00:03:25,900 --> 00:03:28,500
That's it.
They first did not look good.

60
00:03:28,500 --> 00:03:32,100
Like, you're like, okay, it was
like, yeah, but now I feel like

61
00:03:32,100 --> 00:03:33,900
I can't.
I can hardly tell the difference

62
00:03:33,900 --> 00:03:37,100
s so good.
They're getting and that's kind

63
00:03:37,100 --> 00:03:40,000
of the scary part.
Oh no, Kevin.

64
00:03:40,300 --> 00:03:41,700
Am I gonna be able to sleep
tonight?

65
00:03:42,100 --> 00:03:45,900
Well, might, I might disturb you
at the very end.

66
00:03:46,400 --> 00:03:48,800
Fuck great, great.
Great.

67
00:03:48,800 --> 00:03:51,600
Great, great, great.
Hopefully your card in our

68
00:03:51,600 --> 00:03:54,000
future.
So, Yeah, I'm going to talk

69
00:03:54,000 --> 00:03:55,700
about all this stuff.
It's going to be pretty

70
00:03:55,700 --> 00:03:59,900
unbelievable.
And at the end, I think I might

71
00:03:59,900 --> 00:04:06,200
blow your mind in a good way.
You're gonna, it's gonna blow

72
00:04:06,200 --> 00:04:09,900
your mind, okay?
Okay.

73
00:04:09,900 --> 00:04:15,900
I'll just hang on, like like
setbacks of hack, make sure your

74
00:04:15,900 --> 00:04:17,899
seat belts are a nice and
fastened.

75
00:04:17,899 --> 00:04:21,899
Grab one of these have have a
dystopian drink with us.

76
00:04:22,500 --> 00:04:25,000
Yeah.
T.i. hydration, whatever the

77
00:04:25,000 --> 00:04:27,400
fuck it is.
Get rich is not hydration.

78
00:04:27,400 --> 00:04:34,500
Get ready for a ride.
Okay, so first thing we're going

79
00:04:34,500 --> 00:04:40,000
to do wilcott, segment one, and
it's going to be about what our

80
00:04:40,000 --> 00:04:44,400
deep fix.
What really are, they okay, so

81
00:04:44,400 --> 00:04:50,800
defects are artificial media?
That's created using machine

82
00:04:50,800 --> 00:04:54,500
learning techniques such as
things, they Deep neural

83
00:04:54,600 --> 00:04:57,000
networks.
Okay, yeah.

84
00:04:57,400 --> 00:05:01,500
And what they do is they take
the form of manipulated images

85
00:05:01,500 --> 00:05:05,300
or video or audio that appeared
to be real.

86
00:05:05,300 --> 00:05:09,200
But are in fact, purely
synthetic, okay.

87
00:05:09,900 --> 00:05:13,600
Now D pigs are created by
training algorithms with very

88
00:05:13,600 --> 00:05:16,800
large amounts of data, to learn
the characteristics in the

89
00:05:16,800 --> 00:05:21,700
patterns of a particular
individuals, voice face and

90
00:05:21,700 --> 00:05:27,500
mannerisms So it's just studying
videos of that person, for

91
00:05:27,500 --> 00:05:32,000
instance, Tom Cruise, he has all
of those movies and interviews

92
00:05:32,000 --> 00:05:36,200
and precisely.
Yeah, it's just a machine

93
00:05:36,200 --> 00:05:40,300
watching and watching that just
on repeat and just like people

94
00:05:40,300 --> 00:05:42,800
that do Impressions and they
just watch him and sometimes

95
00:05:42,800 --> 00:05:45,400
they're so just so good because
they just watched and studied

96
00:05:45,400 --> 00:05:48,100
them for so long.
These are computers that can be

97
00:05:48,100 --> 00:05:50,000
perfect.
Oh my God.

98
00:05:50,000 --> 00:05:52,400
Kevin.
So once the algorithm has

99
00:05:52,400 --> 00:05:55,800
learned these past Patterns it
can then generate new content

100
00:05:55,800 --> 00:05:59,100
that mimics the individuals
appearance speech and behavior.

101
00:05:59,300 --> 00:06:01,200
Hence, what you were just
talking right of the time Crews

102
00:06:01,200 --> 00:06:07,200
videos now the term deep fake.
Is actually a combination of two

103
00:06:07,200 --> 00:06:12,500
terms, deep learning and fake.
It's reflecting the fact that

104
00:06:12,500 --> 00:06:16,900
these Technologies rely on
complex, algorithms to generate

105
00:06:16,900 --> 00:06:21,100
these convincing simulations of
reality.

106
00:06:22,000 --> 00:06:25,400
Now, defects have been used for
a variety of purposes including

107
00:06:25,700 --> 00:06:30,200
fake news.
All kinds of propaganda revenge

108
00:06:30,200 --> 00:06:32,900
porn.
Oh yeah.

109
00:06:33,500 --> 00:06:36,600
Oh my God, that's awful.
And they can be very difficult

110
00:06:36,600 --> 00:06:40,200
to detect because just by Nature
what they are.

111
00:06:40,200 --> 00:06:45,000
They are designed to look and
sound realistic right now, the

112
00:06:45,000 --> 00:06:48,300
creation and use of deep fakes,
they raised a tremendous number

113
00:06:48,300 --> 00:06:54,700
of ethical and legal concern
again because they can be used

114
00:06:54,700 --> 00:06:57,900
to manipulate public opinion,
they can be used to defame I'm

115
00:06:57,900 --> 00:07:01,400
people individuals, or like we
said, spread disinformation.

116
00:07:02,200 --> 00:07:04,500
Now, as a result of go ahead.
Oh, sorry.

117
00:07:04,500 --> 00:07:07,400
And represent people who.
That's, yeah.

118
00:07:07,400 --> 00:07:10,800
It's not them, right?
Like so exactly like copyrighted

119
00:07:10,800 --> 00:07:13,500
material, right?
Like, like, can you copyright

120
00:07:13,500 --> 00:07:19,200
yourself?
Like okay like let's say let's

121
00:07:19,200 --> 00:07:23,700
say oh God you know what?
I just thought about what we're

122
00:07:23,700 --> 00:07:26,300
doing, we're doing these
podcasts huh?

123
00:07:26,500 --> 00:07:30,900
And we have cameras, yes honest.
People could make deep fakes on

124
00:07:30,900 --> 00:07:33,300
us, just based on are without
question.

125
00:07:34,100 --> 00:07:39,000
Oh my God someone could create a
whole podcast with the two of us

126
00:07:39,100 --> 00:07:41,400
saying sure that we didn't
actually saying should we didn't

127
00:07:41,400 --> 00:07:47,600
actually say and very plausible.
Like in fact, I could go today

128
00:07:47,600 --> 00:07:50,700
and probably do that.
Ah, really?

129
00:07:50,700 --> 00:07:53,600
Uh-huh.
Well Kevin Kevin you are like

130
00:07:53,700 --> 00:07:57,200
tech savvy so like but I'm just
gonna go that's how available it

131
00:07:57,200 --> 00:07:59,800
is.
Like I wouldn't have to like dig

132
00:07:59,800 --> 00:08:01,100
to do.
Figure out how to do that.

133
00:08:01,100 --> 00:08:03,300
I could just go do a couple
searches and figure out how to

134
00:08:03,308 --> 00:08:05,600
do this.
That's fucking crazy.

135
00:08:06,800 --> 00:08:08,700
This is the problem.
What is truth?

136
00:08:08,700 --> 00:08:10,900
What is not?
This is the problem.

137
00:08:10,900 --> 00:08:15,300
Yes, exactly, sorry.
So many governments as a result

138
00:08:15,300 --> 00:08:17,600
of all of this and high-tech
companies are investing in a lot

139
00:08:17,600 --> 00:08:25,200
of research to develop methods
to detect and to prevent The

140
00:08:25,200 --> 00:08:29,000
spread of defects because you
can as based on all of the

141
00:08:29,008 --> 00:08:34,700
things I've just said, it could
be very, very problematic,

142
00:08:35,500 --> 00:08:37,200
right?
Because if you have like a world

143
00:08:37,200 --> 00:08:41,799
leader or something whose like
we just declared nuclear war on

144
00:08:41,799 --> 00:08:45,300
the United States.
Right now we have all of our

145
00:08:45,300 --> 00:08:50,000
Warheads pointing directly to
Canada like who would ever want

146
00:08:50,000 --> 00:08:56,400
to fuck Hannah like Sweden.
But like, but yeah, you could

147
00:08:56,400 --> 00:08:58,800
say that.
And then the whole world was on

148
00:08:58,800 --> 00:09:02,000
all kinds of panic.
Fuck.

149
00:09:02,000 --> 00:09:04,700
So would be in their best
interest to like figure it out?

150
00:09:04,700 --> 00:09:07,000
Yeah.
How can we, how can we determine

151
00:09:07,000 --> 00:09:09,000
whether or not?
This is a deep fake.

152
00:09:09,000 --> 00:09:13,800
Yes.
And then not to throw gas on the

153
00:09:13,800 --> 00:09:19,700
fire but then to determine How
do you determine the people that

154
00:09:19,700 --> 00:09:23,100
are saying it's fake when it's
not fake right?

155
00:09:23,300 --> 00:09:27,200
Because they could be on video
or even a that you've just you

156
00:09:27,200 --> 00:09:29,200
just made that I didn't.
I wouldn't have done that.

157
00:09:29,200 --> 00:09:31,200
I didn't take my pants down for
that prostitute.

158
00:09:31,600 --> 00:09:34,900
Oh, yeah.
Everything could be a deep fake,

159
00:09:34,900 --> 00:09:37,500
anything.
I mean it's a convenient excuse

160
00:09:37,500 --> 00:09:42,000
for audio video, pictures shit.
Yeah, that's true.

161
00:09:42,900 --> 00:09:45,800
So What are the types of
defects?

162
00:09:45,800 --> 00:09:49,100
I just said images videos.
Pictures.

163
00:09:49,300 --> 00:09:53,000
Yeah.
So in 2019 sounds yes.

164
00:09:53,400 --> 00:09:57,200
So in 2019, a date, a deep fake
video of former President.

165
00:09:57,200 --> 00:09:59,900
Barack Obama was created, do you
remember this?

166
00:09:59,900 --> 00:10:03,500
That's the first one that came
into my mind where he appeared

167
00:10:03,500 --> 00:10:06,300
to give a speech that he never
ever gave.

168
00:10:06,300 --> 00:10:09,300
And the video was created by
training an algorithm on

169
00:10:09,300 --> 00:10:12,600
existing footage of Mama and in
manipulating it to look like he

170
00:10:12,600 --> 00:10:14,000
was saying something that he
never said.

171
00:10:14,100 --> 00:10:16,200
Said that's dangerous.
Scary.

172
00:10:16,300 --> 00:10:18,300
Yeah.
So the video was created as a

173
00:10:18,308 --> 00:10:23,400
demonstration.
Of the potential dangers of

174
00:10:23,400 --> 00:10:27,700
defects.
It was to prove a point and show

175
00:10:27,700 --> 00:10:31,300
just how easy someone could
create that video and spread up

176
00:10:31,400 --> 00:10:36,000
false information on a dime.
The video went quickly viral.

177
00:10:36,000 --> 00:10:39,200
Yeah, sparked.
A lots of discussions about deep

178
00:10:39,200 --> 00:10:46,300
fakes on politics and Society,
it was cray cray and we're going

179
00:10:46,300 --> 00:10:48,100
to watch it.
We're gonna watch it.

180
00:10:48,100 --> 00:10:49,700
Okay, great?
I because I would like to see it

181
00:10:49,700 --> 00:10:51,500
again.
Here we go.

182
00:10:51,700 --> 00:10:55,000
Okay.
We're entering an era in which

183
00:10:55,000 --> 00:10:57,900
our enemies can make it look.
Like I didn't wanna say anything

184
00:10:57,900 --> 00:11:00,900
at any point in time, even if
they would never say those

185
00:11:00,900 --> 00:11:05,000
things.
So for instance, they could have

186
00:11:05,000 --> 00:11:08,100
me say things like, I don't
know.

187
00:11:08,800 --> 00:11:13,300
Kill mother was right.
Or Ben Carson is in the second

188
00:11:13,300 --> 00:11:18,200
place or how about the simply
President Trump is a total and

189
00:11:18,200 --> 00:11:22,000
complete edition?
Well I mean strength see I would

190
00:11:22,000 --> 00:11:25,700
never say that.
At least not a public address,

191
00:11:25,700 --> 00:11:31,300
but someone else would someone
like Jordan Peele?

192
00:11:32,800 --> 00:11:36,800
This is dangerous time.
Moving forward.

193
00:11:36,800 --> 00:11:39,900
We need to be more Vigilant with
what we trust from the internet.

194
00:11:40,600 --> 00:11:42,900
That's the time we need to rely
on.

195
00:11:42,900 --> 00:11:50,600
Trusted news source, may sound
basic but how are we for age of

196
00:11:50,600 --> 00:11:54,800
information is going to be the
difference between whether we

197
00:11:54,800 --> 00:11:58,300
survive or whether we become
some kind of fucked up to

198
00:11:58,308 --> 00:12:02,300
system.
Thank you stable, bitches.

199
00:12:06,000 --> 00:12:11,900
So, if you're watching on video,
did you notice how the Obama

200
00:12:11,900 --> 00:12:16,000
version the mannerisms were
completely different?

201
00:12:16,000 --> 00:12:18,100
Yes.
Then what Jordan was doing?

202
00:12:18,100 --> 00:12:20,100
Yeah.
Because the computer was

203
00:12:20,100 --> 00:12:23,000
generating those movements based
on what he was saying, and the

204
00:12:23,000 --> 00:12:28,100
based on the movements that
Obama makes when he talks, you

205
00:12:28,100 --> 00:12:31,600
know what I kept, the mouth
looked a little to me.

206
00:12:31,600 --> 00:12:34,100
I could see a little something
off, on the little something off

207
00:12:34,100 --> 00:12:35,500
on the mouth, that's what I
wrote.

208
00:12:35,700 --> 00:12:38,100
Down mouth.
It was kind of fuzzy, like, the

209
00:12:38,100 --> 00:12:42,700
mouse kind of looked like, it
was like, moved out a little

210
00:12:42,700 --> 00:12:43,500
bit.
Yeah.

211
00:12:43,700 --> 00:12:46,000
And it wasn't right exact and
like that.

212
00:12:46,000 --> 00:12:48,000
Like, the dimensions of Obama's
face.

213
00:12:48,100 --> 00:12:50,300
Yes.
But I did, when I was listening

214
00:12:50,300 --> 00:12:52,800
to her, I thought this kind of
sounds like Obama, but it really

215
00:12:52,800 --> 00:12:54,600
doesn't sound like Obama, right?
Right?

216
00:12:54,600 --> 00:12:56,700
Which so here's the here's the
thing.

217
00:12:56,800 --> 00:13:00,600
Okay, two things.
So in your first point on the

218
00:13:01,500 --> 00:13:07,300
visuals, yeah, In from the time
that that was made until today

219
00:13:07,600 --> 00:13:10,200
the technology has improved 20
times.

220
00:13:10,200 --> 00:13:14,700
What year was that again?
19 I believe 1920 19 so in four

221
00:13:14,700 --> 00:13:19,300
years it has drastically gotten
better Goddamn.

222
00:13:19,400 --> 00:13:23,000
Secondly he was he just has an
impersonator sweat, it has voice

223
00:13:23,000 --> 00:13:25,500
but now they have synthetic
generators.

224
00:13:26,200 --> 00:13:28,600
Yeah we'll do it.
You don't have to actually

225
00:13:28,900 --> 00:13:31,900
pretend to be them so then you
really do get the tones.

226
00:13:32,200 --> 00:13:38,800
I I'm so mad at myself.
Had went to a website and I had

227
00:13:38,800 --> 00:13:44,300
it train my voice and I had it
speaking things on my behalf and

228
00:13:44,300 --> 00:13:46,900
I couldn't find that link
anymore so I didn't have that

229
00:13:46,900 --> 00:13:50,400
particular clip but it sounded
just like me.

230
00:13:50,800 --> 00:13:52,800
Even your intonations in my
intonations.

231
00:13:52,800 --> 00:13:56,000
The way I said, the words, you
know, what's crazy about that.

232
00:13:56,000 --> 00:14:00,900
They also know how filters like
current filters icon Tick, Tock

233
00:14:00,900 --> 00:14:04,200
where someone could be narrating
things and it's like it's almost

234
00:14:04,300 --> 00:14:07,900
Almost like an effect just on
the sound, but it sounds like a

235
00:14:07,908 --> 00:14:10,700
totally different person person.
Like, maybe it would sound like

236
00:14:10,700 --> 00:14:17,900
your voice or like see and
that's, that's clearly trained

237
00:14:17,900 --> 00:14:19,800
on someone else's voice.
Yes.

238
00:14:20,400 --> 00:14:24,700
What the fuck like, okay, also
can we just talk about how

239
00:14:24,700 --> 00:14:29,000
awesome Jordan Peele is because
he is he's not a friend of the

240
00:14:29,000 --> 00:14:31,900
Pod but I would love for him to
be a friend of the Pod because

241
00:14:31,900 --> 00:14:35,500
he is a fellow like he's Do
enthusiasts?

242
00:14:35,500 --> 00:14:38,200
Yes.
Yeah, we're jobs the crazy shit,

243
00:14:38,200 --> 00:14:40,400
right?
Yeah, you woke like, honestly,

244
00:14:40,800 --> 00:14:44,600
like stay vigilant on false
information.

245
00:14:44,600 --> 00:14:47,600
Yes.
That what's crazy is like when

246
00:14:47,600 --> 00:14:50,200
that video came out?
Yeah, it went insane.

247
00:14:50,200 --> 00:14:53,200
Yeah and everyone was like, oh
my God, what else can we make?

248
00:14:53,200 --> 00:14:55,000
Obama say.
Like that's crazy.

249
00:14:55,000 --> 00:14:57,700
Yes and now we look at it, we're
like okay, our eyes are kind of

250
00:14:57,700 --> 00:14:59,300
train.
Yes.

251
00:14:59,300 --> 00:15:02,000
So we kind of can see that
they're like I said the

252
00:15:02,000 --> 00:15:04,200
technology is just getting way
way.

253
00:15:04,300 --> 00:15:06,100
Way better?
Yeah.

254
00:15:07,800 --> 00:15:09,800
Oh my God.
It makes me so nervous.

255
00:15:10,500 --> 00:15:15,200
Great has great.
So photos are probably the

256
00:15:15,200 --> 00:15:17,400
easiest to fake because we've
actually been doing that for a

257
00:15:17,400 --> 00:15:20,100
very long time.
Hello, Photoshop.

258
00:15:20,200 --> 00:15:24,400
Yeah, facetune name.
Anything, even back in the day,

259
00:15:24,400 --> 00:15:27,800
like 1940s, they used to put
like Vaseline, and like they

260
00:15:27,800 --> 00:15:30,900
would even go into the actual
negatives and like scraping,

261
00:15:30,900 --> 00:15:37,200
yeah, it's always been a thing
to A the thing about those

262
00:15:37,200 --> 00:15:39,600
filters are like the filters and
things like that.

263
00:15:39,800 --> 00:15:43,000
Often times, you can't tell the
difference, like there are some

264
00:15:44,500 --> 00:15:48,400
some accounts that I follow that
show how people have edited

265
00:15:48,400 --> 00:15:51,800
their, their bodies or their
faces like even people like

266
00:15:51,800 --> 00:15:53,800
Naomi Campbell okay.
Yeah.

267
00:15:53,800 --> 00:15:55,900
Who's a supermodel?
Beautiful?

268
00:15:55,900 --> 00:15:58,700
Like ridiculous trying to make
everything, perfect.

269
00:15:58,800 --> 00:16:03,800
And she's like, changing her
like neck to be longer like her

270
00:16:03,800 --> 00:16:07,300
face to To be thinner.
Yes or like, you know, wrinkles

271
00:16:07,300 --> 00:16:09,800
to go away.
And I can't see the, I didn't

272
00:16:09,800 --> 00:16:13,000
know, but there are some,
there's one called celeb face.

273
00:16:13,000 --> 00:16:16,100
I don't know if you've seen that
one, but they'll show like, the

274
00:16:16,300 --> 00:16:18,800
picture that the original
picture and the picture that the

275
00:16:18,800 --> 00:16:23,400
person posted and show, like,
it'll show the original picture

276
00:16:23,400 --> 00:16:25,800
or no first, we'll show the
picture of the person posted

277
00:16:25,800 --> 00:16:26,900
like, I don't know, Kylie
Jenner.

278
00:16:26,900 --> 00:16:29,300
Sure.
The next, the next picture will

279
00:16:29,300 --> 00:16:32,700
be the original picture.
Then they'll show in a video

280
00:16:32,700 --> 00:16:36,900
clip the edits that were made.
So it looks like a transform it.

281
00:16:36,900 --> 00:16:40,100
Yeah, to transform it and you
see you're like whoa, they don't

282
00:16:40,100 --> 00:16:42,400
even look like the same person,
right?

283
00:16:42,700 --> 00:16:46,000
Like people are almost making
themselves look like fucking

284
00:16:46,000 --> 00:16:48,800
like aliens.
I like some kind of weird

285
00:16:48,800 --> 00:16:51,200
represent like false
representation of who they are

286
00:16:51,200 --> 00:16:53,600
as a person.
Well, even those very, very,

287
00:16:53,600 --> 00:16:56,500
very basic simple ones.
Like the do makeup.

288
00:16:56,600 --> 00:16:58,400
Yes, looks so real.
Yeah.

289
00:16:58,400 --> 00:17:02,100
The makeup filters.
Yes, I'm like, I would have bet

290
00:17:02,100 --> 00:17:04,500
a million dollars that you were
wearing that makeup.

291
00:17:04,500 --> 00:17:06,900
Yes.
Yeah, it's yeah.

292
00:17:07,400 --> 00:17:11,000
There's one, there's one right
now on Tick-Tock that like

293
00:17:11,000 --> 00:17:14,900
everyone's using and they, you
know, the Hunchback of Notre

294
00:17:14,900 --> 00:17:16,200
Dame.
Oh you don't really like

295
00:17:16,200 --> 00:17:17,800
musicals.
That's right.

296
00:17:18,200 --> 00:17:24,099
But there's a song where the
this guy saying you are deformed

297
00:17:24,099 --> 00:17:27,900
and then Quasimodo's like I am
deformed so people are taking

298
00:17:27,900 --> 00:17:31,500
the filter off of them so that
the I'm saying I am deformed

299
00:17:31,500 --> 00:17:34,400
either Arisen.
Then you you're like Quasimodo

300
00:17:34,700 --> 00:17:37,400
like your actual picture like
who you are.

301
00:17:37,400 --> 00:17:39,700
Yeah.
And it's like it does look

302
00:17:39,700 --> 00:17:42,800
deformed.
It's like this weird like false

303
00:17:42,800 --> 00:17:45,100
representation of like you on
your best day with like a

304
00:17:45,100 --> 00:17:48,700
shitload of fucking plastic
surgery and stuff like that and

305
00:17:48,700 --> 00:17:50,400
stuff and makeup professionally
done.

306
00:17:50,400 --> 00:17:52,000
Yeah.
But that you can't even tell

307
00:17:52,000 --> 00:17:53,800
that they're even wearing
filters anymore.

308
00:17:53,900 --> 00:17:55,400
Are we going to know what's real
anymore?

309
00:17:55,800 --> 00:17:59,900
That's what I'm what I'm
wondering is even years ago.

310
00:17:59,900 --> 00:18:03,500
Bobby Brown, you know, Bobby not
Bobby Brown, the singer Bobby

311
00:18:03,500 --> 00:18:04,900
Brown, the makeup artist.
Yes.

312
00:18:05,300 --> 00:18:10,200
So she caught a lot of shit
because she was using a filter

313
00:18:10,200 --> 00:18:13,100
on her videos to make her look
younger.

314
00:18:13,100 --> 00:18:14,900
And to make her look more
radiant.

315
00:18:15,100 --> 00:18:17,300
Not disclose marketing.
That's some bad.

316
00:18:17,300 --> 00:18:20,400
That's misleading.
Yeah, because you're doing a

317
00:18:20,400 --> 00:18:22,900
makeup brand, yes, right.
And you're saying, look how

318
00:18:22,900 --> 00:18:27,000
beautiful I am when you're, in
fact, you're doing technological

319
00:18:27,200 --> 00:18:28,800
editing.
That's yes, that's not right?

320
00:18:28,800 --> 00:18:31,900
It's false representation of
False representation, totally,

321
00:18:31,900 --> 00:18:33,700
which is interesting.
I don't, I don't think we will

322
00:18:33,700 --> 00:18:36,100
be able to tell, I don't either,
I don't either.

323
00:18:37,800 --> 00:18:45,400
So, With a is software, we don't
need Photoshop anymore.

324
00:18:45,400 --> 00:18:46,700
You have to be good at
Photoshop.

325
00:18:46,700 --> 00:18:49,800
Like Photoshop is a chore.
It's not like you need to know

326
00:18:49,800 --> 00:18:50,800
what?
Yeah, it's an art.

327
00:18:50,800 --> 00:18:54,500
You have to know what you're
doing but with AI you just kind

328
00:18:54,500 --> 00:18:57,700
of go click like, you did.
Oh, I want to look pretty push

329
00:18:57,700 --> 00:18:59,300
the button Loop.
All the come pretty.

330
00:18:59,300 --> 00:19:01,400
I mean, it.
Yes, at your fingertips,

331
00:19:01,400 --> 00:19:09,400
literally. so I took a photo of
myself and a photo of Alicia and

332
00:19:09,400 --> 00:19:15,800
I put it in a software called
reface, And I asked it to put us

333
00:19:15,800 --> 00:19:23,800
on the cover of a Playboy.
And in about 20 seconds poof we

334
00:19:23,800 --> 00:19:30,500
were on Playboy as one of my
favorite photos of all time and

335
00:19:30,500 --> 00:19:34,000
there it is.
Look at my mouth.

336
00:19:34,000 --> 00:19:36,300
Okay girl.
Yeah you're sexy with that

337
00:19:36,300 --> 00:19:38,600
little pouty.
Mouth and your lond hair and

338
00:19:38,600 --> 00:19:42,200
you're right firm bosoms.
Yes.

339
00:19:42,200 --> 00:19:43,200
Oh my God.
Yeah.

340
00:19:43,200 --> 00:19:44,900
That girl doesn't need to wear a
bra.

341
00:19:45,000 --> 00:19:49,800
No, look at you like yes.
Yes, you might ease Farah hair.

342
00:19:51,500 --> 00:19:54,700
You may have like a little bit
of a mustache.

343
00:19:54,700 --> 00:19:59,300
I do have a mustache happening.
So I'm a little like, what is it

344
00:20:00,500 --> 00:20:03,900
non-transforming or non gender
gender?

345
00:20:05,400 --> 00:20:07,000
What is that?
What is that?

346
00:20:07,000 --> 00:20:09,000
Yeah.
You know, I'm male and female

347
00:20:09,000 --> 00:20:11,300
and tragic and drawed.
Yeah, it's very androgynous

348
00:20:11,300 --> 00:20:12,200
picture.
Yeah.

349
00:20:12,800 --> 00:20:14,100
Oh, my isn't it?
Hot.

350
00:20:14,200 --> 00:20:16,700
That is so funny.
I can see my eyes for sure.

351
00:20:16,700 --> 00:20:20,200
And I guess I can see my nose.
Yeah, I I guess I kind of like,

352
00:20:20,800 --> 00:20:22,800
My mouth.
I think my mouth your mouth is a

353
00:20:22,800 --> 00:20:24,700
little different.
Yeah, yeah, pouty.

354
00:20:25,100 --> 00:20:30,500
Hmm.
I Cans I definitely can see,

355
00:20:30,500 --> 00:20:32,600
it's definitely Raw.
It's definitely a definitely my

356
00:20:32,600 --> 00:20:35,700
eyes and definitely like that's
definitely a that like, that's

357
00:20:35,700 --> 00:20:39,100
fucking hilarious.
Playmate sister.

358
00:20:43,200 --> 00:20:45,300
So, you didn't even say like,
which one?

359
00:20:45,400 --> 00:20:47,100
No.
You just said, take these two

360
00:20:47,100 --> 00:20:49,100
pictures.
Put us on a Playboy, cut it.

361
00:20:50,500 --> 00:20:53,200
God.
Dude, that is some crazy shit

362
00:20:53,200 --> 00:20:54,500
crazy.
Yeah.

363
00:20:55,600 --> 00:20:57,900
Crazy.
And that's, that's, that's like

364
00:20:58,200 --> 00:21:00,400
20 seconds cursory glance.
Yes.

365
00:21:00,400 --> 00:21:03,400
I am sure.
You can find something that you

366
00:21:03,400 --> 00:21:06,200
spend money on this.
Like, you know.

367
00:21:06,200 --> 00:21:07,800
Yeah.
Like intense.

368
00:21:07,800 --> 00:21:10,800
And like it takes processing,
like you imagine.

369
00:21:12,000 --> 00:21:14,400
I can't go.
That's fucking free 22nd thing.

370
00:21:14,400 --> 00:21:16,200
Yeah, exactly.
That's my point.

371
00:21:16,200 --> 00:21:23,300
That is my point.
So, The point.

372
00:21:23,800 --> 00:21:25,500
Yeah, I guess that is.
That was my point of that.

373
00:21:25,500 --> 00:21:27,500
Picture is how simple?
Yeah.

374
00:21:27,500 --> 00:21:31,800
Easy and accessible it is.
And to your point, how you can

375
00:21:31,800 --> 00:21:34,500
imagine?
There is stuff that exists that

376
00:21:34,500 --> 00:21:38,400
takes way more processing power
would cost me, you know, take

377
00:21:38,400 --> 00:21:40,700
some time and money to make it
so you can time.

378
00:21:40,700 --> 00:21:46,900
You can imagine how unknown.
How realistic so realistic that

379
00:21:46,900 --> 00:21:49,300
you couldn't tell the difference
like you literally just played

380
00:21:49,300 --> 00:21:52,400
you just posed for Playboy.
Didn't you right?

381
00:21:52,400 --> 00:21:54,100
Right?
What if okay, what if you're

382
00:21:54,500 --> 00:21:56,100
okay?
What if you're getting a new job

383
00:21:56,100 --> 00:21:57,400
or something?
Exactly.

384
00:21:57,400 --> 00:22:00,600
And these post like these
pictures are our poster of you

385
00:22:00,600 --> 00:22:02,500
like you're saying a revenge
porn and stuff.

386
00:22:02,500 --> 00:22:04,100
It's like it would have someone
takes that picture.

387
00:22:04,100 --> 00:22:07,400
I just put online they take a
snapshot of it.

388
00:22:07,500 --> 00:22:11,200
Publish it online and go Kevin
was doing Playboy's a trans.

389
00:22:11,200 --> 00:22:13,400
Yeah.
I mean they could make anything

390
00:22:13,600 --> 00:22:19,000
You know I'm saying?
Yeah, it's like I said good and

391
00:22:19,000 --> 00:22:21,400
evil.
Yeah, good and evil.

392
00:22:21,400 --> 00:22:23,200
I think it's good because I
think it's hot.

393
00:22:25,300 --> 00:22:28,300
You look hot, great, like
nobody's making a filter to make

394
00:22:28,300 --> 00:22:32,800
you look uglier than exactly.
Yes, we're fine.

395
00:22:32,800 --> 00:22:35,800
With we're fine with a Playboy
picture, like just don't

396
00:22:35,800 --> 00:22:37,700
superimpose our faces on a
Manitou years.

397
00:22:37,700 --> 00:22:43,600
Exactly.
So in 2021, that audio device, A

398
00:22:43,608 --> 00:22:45,600
pic of Tom Cruise that you were
talking about.

399
00:22:45,600 --> 00:22:48,600
Yes, that was created by a
company called Reese feature.

400
00:22:49,800 --> 00:22:54,200
In any audio clip, a person in
impersonating, Tom was recorded

401
00:22:54,200 --> 00:22:55,700
saying things that he never
said.

402
00:22:56,000 --> 00:22:59,500
And the clip was created by
analyzing existing recordings of

403
00:22:59,500 --> 00:23:03,400
Tom Cruise and then it's
synthesized brand new words and

404
00:23:03,400 --> 00:23:06,900
phrases using these machine
learning algorithms.

405
00:23:07,900 --> 00:23:11,300
Now, the Deep fake audio clip
was so convincing that it fools

406
00:23:11,300 --> 00:23:13,600
so many people into thinking,
Tom was actually doing it.

407
00:23:13,600 --> 00:23:17,100
I mean, it took a hot minute for
people to really think that it

408
00:23:17,108 --> 00:23:19,200
wasn't him being cray.
Cray, I thought the same thing,

409
00:23:19,400 --> 00:23:21,300
Yeah, exactly because he has
been cray-cray.

410
00:23:21,300 --> 00:23:26,700
Well that's that's just it.
It was hard but like extreme

411
00:23:26,700 --> 00:23:28,600
extreme personality.
Which is great.

412
00:23:28,600 --> 00:23:30,500
We love those very interesting.
Yeah.

413
00:23:30,500 --> 00:23:34,000
But not out of the world out of
this.

414
00:23:34,000 --> 00:23:36,700
So that that is, you know, the
clip that really started raising

415
00:23:36,700 --> 00:23:39,900
concerns to about the potential
measures of deep fake audio

416
00:23:39,900 --> 00:23:44,000
technology because that's even
worse because you can't see the

417
00:23:44,200 --> 00:23:46,700
mouth, moving, weird or
whatever.

418
00:23:46,700 --> 00:23:49,600
Could just be some kind of like
audio clip released which Which

419
00:23:49,600 --> 00:23:51,600
does happen.
And it happened on many

420
00:23:51,600 --> 00:23:55,000
occasions.
What Mel Gibson remember when he

421
00:23:55,000 --> 00:23:57,200
came out like?
Oh, yeah, but those like

422
00:23:57,400 --> 00:23:59,000
voicemails and stuff.
Yes.

423
00:23:59,100 --> 00:24:01,800
And you're like, damn, you like,
oh yeah, you're like wait a

424
00:24:01,808 --> 00:24:07,400
second, you're racist homophobic
like piece of shit.

425
00:24:07,400 --> 00:24:08,400
Uh-huh.
Cool.

426
00:24:08,400 --> 00:24:10,700
And it cost him his entire
career head.

427
00:24:11,200 --> 00:24:13,000
What if it did?
What if that was fake?

428
00:24:14,300 --> 00:24:17,800
I mean, I mean, no, I don't
think it was, but what if I was?

429
00:24:17,800 --> 00:24:20,700
So what if it was?
Or what if what if somebody has

430
00:24:20,700 --> 00:24:23,200
kind of some kind of a vendetta
against show?

431
00:24:23,400 --> 00:24:26,000
Exactly.
And can use, I don't know,

432
00:24:26,100 --> 00:24:31,200
podcast Clips to to make you say
when your career real life.

433
00:24:32,100 --> 00:24:33,800
Yeah.
How do you protect that?

434
00:24:34,700 --> 00:24:38,000
This is what people need.
This is what governments and

435
00:24:38,400 --> 00:24:39,900
we're going to talk about in a
second but something that needs

436
00:24:39,900 --> 00:24:41,300
to be figured out.
Yeah.

437
00:24:44,300 --> 00:24:48,800
So, you know, it's worth noting
that while the audio deep fakes

438
00:24:48,800 --> 00:24:51,800
are less common than the video.
Deep fakes they are, still a

439
00:24:51,808 --> 00:24:55,100
growing concern and reach users
researchers are working on,

440
00:24:55,100 --> 00:24:58,600
developing new methods to detect
and prevent the spread of audio

441
00:24:58,600 --> 00:25:00,800
deep Vegas as well.
That's that's the thing

442
00:25:00,800 --> 00:25:05,900
happening.
So, I wanted to give this a try.

443
00:25:07,300 --> 00:25:10,800
So, I randomly picked a site
that claims, to be able to do

444
00:25:10,900 --> 00:25:14,600
this activity, and it's called
fake you.com.

445
00:25:16,300 --> 00:25:19,300
And it appears that these voices
are built by community members,

446
00:25:19,300 --> 00:25:21,200
like, people that are really
geeked out on this.

447
00:25:21,400 --> 00:25:25,400
They actually develop these
voices through training and and

448
00:25:25,400 --> 00:25:28,000
such and then they make them
public for people to be able to

449
00:25:28,000 --> 00:25:32,600
use.
Yeah, so so some are good We're

450
00:25:32,600 --> 00:25:35,900
not, okay?
And it just depends on the

451
00:25:35,900 --> 00:25:40,900
amount of time that they or the
AI that they're using put, they

452
00:25:40,900 --> 00:25:45,200
put into it and it takes time
for these systems to to learn.

453
00:25:47,500 --> 00:25:50,500
But also, as an interesting
fact, most if not all the

454
00:25:50,500 --> 00:25:56,000
voices, produced by this website
are actually a Woman by the name

455
00:25:56,000 --> 00:26:01,600
of Linda Johnson with some
layers on top of it now, There's

456
00:26:01,600 --> 00:26:06,200
this thing called a Linda
Johnson speech data set that is

457
00:26:06,200 --> 00:26:11,500
available free on GitHub what?
And apparently, her voice is the

458
00:26:11,500 --> 00:26:14,500
root source of a lot of these
deep fakes and maybe there's

459
00:26:14,500 --> 00:26:18,800
just something about her voice
that allows it to be this kind

460
00:26:18,800 --> 00:26:23,300
of root Source somehow and then
with layers and tones and stuff

461
00:26:23,300 --> 00:26:26,000
they can build just about
anything and she's I guess she's

462
00:26:26,000 --> 00:26:29,200
a kind of a professional
voice-over person.

463
00:26:29,200 --> 00:26:30,900
We're going to take a little
listen.

464
00:26:31,200 --> 00:26:33,600
To a sample of her voice.
Yeah.

465
00:26:35,400 --> 00:26:40,000
Sorry, I can't.
Many animals of even complex

466
00:26:40,000 --> 00:26:43,400
structure which live
parasitically within others are

467
00:26:43,400 --> 00:26:46,000
wholly.
Devoid of an elementary cavity.

468
00:26:47,300 --> 00:26:50,600
So that's Linda Johnson, and
that's a sample.

469
00:26:50,800 --> 00:26:54,300
That's a sample of her data set,
okay?

470
00:26:55,100 --> 00:26:59,300
And so people, I guess somehow
take that, I don't claim to

471
00:26:59,300 --> 00:27:03,100
understand any of this and and
generate these voices.

472
00:27:03,100 --> 00:27:07,100
Now, my first attempt was with a
voice, we all know a mr.

473
00:27:07,100 --> 00:27:08,600
Tom Cruise.
Yes.

474
00:27:08,600 --> 00:27:13,500
Okay, okay, and now it's marked
the system marked its voice as

475
00:27:14,200 --> 00:27:17,700
voice sounds man.
That's literally.

476
00:27:17,700 --> 00:27:19,300
What is it voice?
Sounds Me.

477
00:27:20,500 --> 00:27:26,100
Maybe not fancy.
All right, so I asked it to say

478
00:27:27,000 --> 00:27:28,200
hi.
This is Tom.

479
00:27:28,300 --> 00:27:31,400
One of the things I love more
than Scientology is the bizarre

480
00:27:31,400 --> 00:27:39,900
AF shut.
Oh my God this is what it sounds

481
00:27:39,900 --> 00:27:42,800
like.
Hey, this is Tom.

482
00:27:42,800 --> 00:27:45,600
The one thing I love more than
skin will be is the bizarre of

483
00:27:46,600 --> 00:27:49,400
now observers of incentives.
Yeah, yeah.

484
00:27:49,400 --> 00:27:51,500
First of all, didn't pronounce
words, right?

485
00:27:51,600 --> 00:27:53,600
You're right here the digital
tone of.

486
00:27:53,600 --> 00:27:58,100
Yes, it's you could hear me.
Homs crew Crews tone in that.

487
00:27:58,100 --> 00:28:00,000
Yeah.
Like At first I was like, oh, I

488
00:28:00,000 --> 00:28:02,300
don't know about this Tin.
Can ya sound like a tin can?

489
00:28:02,300 --> 00:28:04,600
But then like, as I was
listening like oh, I can

490
00:28:04,600 --> 00:28:06,800
actually hear it.
Yeah, this is one but just

491
00:28:06,800 --> 00:28:09,300
listen one more time.
Hey, this is Tom.

492
00:28:09,300 --> 00:28:11,100
The one thing I love more than
skin glow guy.

493
00:28:11,300 --> 00:28:13,400
Is the busier of so you can
hear.

494
00:28:13,500 --> 00:28:14,700
Yeah.
And I think that's probably why

495
00:28:14,700 --> 00:28:17,100
they went Sounders me.
Yeah.

496
00:28:18,000 --> 00:28:20,600
It was great.
So that one needs some work.

497
00:28:20,600 --> 00:28:25,800
So next, I tried Barack.
See if the guy that did the

498
00:28:25,800 --> 00:28:29,900
Baroque tone was any better?
Yes, right.

499
00:28:30,200 --> 00:28:34,900
And here's and I had time I had
him say hi.

500
00:28:34,900 --> 00:28:38,700
This is Barack listen to the
bizarre AF because crazy shit is

501
00:28:38,700 --> 00:28:42,600
good shit.
This is what this is.

502
00:28:42,600 --> 00:28:45,300
Brock.
Hello citizens of the United

503
00:28:45,300 --> 00:28:47,300
States.
This is Mark, listen to the

504
00:28:47,300 --> 00:28:50,200
bazzara because crazy shit is
good shit.

505
00:28:51,300 --> 00:28:55,000
It's getting better and better.
Yeah, still still kind of tin

506
00:28:55,000 --> 00:28:57,800
ear.
The Tammy little not saying AF

507
00:28:57,800 --> 00:29:01,100
he's going if yeah and you could
even say his name, right?

508
00:29:01,100 --> 00:29:03,400
You can use a bra crackling.
Yeah.

509
00:29:03,700 --> 00:29:05,600
So that obviously needed some
work.

510
00:29:06,400 --> 00:29:08,500
Okay, couldn't stop.
Was it getting addictive?

511
00:29:08,500 --> 00:29:15,900
Right, I can only see it.
Try this shit so then I'm like I

512
00:29:15,900 --> 00:29:19,600
found Snoop Dogg.
Oh my God I love Snoop Snoopy.

513
00:29:20,700 --> 00:29:23,100
This one, I don't even need to
tell you what he says, check

514
00:29:23,100 --> 00:29:25,600
this out.
Don't fiddle with the shizzle.

515
00:29:25,800 --> 00:29:30,200
Listen to the but so a if that's
pretty good, I'm good.

516
00:29:30,300 --> 00:29:34,600
Yeah, still can't say AF.
Yes it kept till could hear some

517
00:29:35,200 --> 00:29:37,200
synthesis.
Yeah, I'm 10.

518
00:29:37,200 --> 00:29:39,500
Yeah.
The 10 way, better way.

519
00:29:39,500 --> 00:29:41,700
Better best so far best.
So far.

520
00:29:41,700 --> 00:29:47,300
Yeah.
So again I'm like I just I need

521
00:29:47,300 --> 00:29:51,200
to find a good one and then I
found Bart Simpson.

522
00:29:51,400 --> 00:29:55,000
Mmm.
We had guided them under some

523
00:29:55,000 --> 00:29:57,000
Bart.
Yeah, Bart because Doster Dom

524
00:29:57,000 --> 00:30:01,000
estradiol like he belongs in our
fucking bizarre EF world right

525
00:30:01,800 --> 00:30:06,100
now.
He's, he's ok.

526
00:30:06,100 --> 00:30:08,400
But OK, You're Gonna Hear Bart.
Okay.

527
00:30:08,600 --> 00:30:11,800
Okay, here we go.
Here we go, Bart.

528
00:30:11,800 --> 00:30:15,300
Give it to us to do.
The bisseroff, is my sister

529
00:30:15,300 --> 00:30:19,000
Lisa's favorite podcast.
She's a brilliant standest scan

530
00:30:19,000 --> 00:30:21,400
test.
Scant is supposed to say

531
00:30:21,400 --> 00:30:25,700
scientist.
Yeah, so still having

532
00:30:25,700 --> 00:30:30,500
pronunciation problem.
Yes, sir off is my sister Lisa's

533
00:30:30,500 --> 00:30:33,300
favorite podcast.
She's a brilliant standest.

534
00:30:34,200 --> 00:30:37,300
But like, even like it's
sounding like the intonations

535
00:30:37,300 --> 00:30:39,000
our sound better.
Yeah.

536
00:30:39,000 --> 00:30:43,700
And it, although, it still
sounds tinny it.

537
00:30:43,800 --> 00:30:47,300
I mean, it's way better.
Yes, it's way better way better.

538
00:30:47,400 --> 00:30:48,800
And remember, keep in mind, this
is free.

539
00:30:49,000 --> 00:30:52,300
I'll free options.
Free that you just put yourself

540
00:30:52,300 --> 00:30:55,000
into a queue, it'll generate it
and then he says I'm done.

541
00:30:55,500 --> 00:30:57,900
So again keep him.
Keep that in mind when we're

542
00:30:57,900 --> 00:31:00,600
doing that.
And oh, look who just walked in

543
00:31:00,600 --> 00:31:04,800
the studio.
So, Gwyneth Paltrow just walked

544
00:31:04,800 --> 00:31:07,600
into the studio.
And for those that are on video,

545
00:31:07,600 --> 00:31:09,600
it's she's off Mikey just you
can't see her.

546
00:31:09,600 --> 00:31:12,700
She's kind of right over here,
but the producers in my ear just

547
00:31:12,700 --> 00:31:15,200
told me.
She's she was in the lot and she

548
00:31:15,200 --> 00:31:17,800
just wanted to say hi.
Hey Gwyneth, how you doing?

549
00:31:17,800 --> 00:31:22,000
Hey, how's it going?
Wait, where are you there?

550
00:31:22,000 --> 00:31:23,700
Thanks for the introduction.
Chemical.

551
00:31:23,700 --> 00:31:25,500
Sure.
As you said, I'm going to and

552
00:31:25,500 --> 00:31:28,200
I'm so happy to be here.
I've listened to all of your

553
00:31:28,200 --> 00:31:32,100
podcasts and you guys are great.
Kevin, you are the hottest guy I

554
00:31:32,100 --> 00:31:36,300
have ever seen and Alicia love
your Playboy photo and I'd love

555
00:31:36,300 --> 00:31:39,100
for you to do a group photo
shoot for our next cover.

556
00:31:39,200 --> 00:31:43,000
How nice I would love to hear.
So sweet that.

557
00:31:43,000 --> 00:31:44,700
Oh you got it.
Okay, no problem.

558
00:31:44,700 --> 00:31:47,600
Thanks for stopping by.
It's a can you leave us a candle

559
00:31:47,600 --> 00:31:49,800
or something?
Yeah smells like your the vag

560
00:31:49,800 --> 00:31:51,900
badge egg.
I just wanted to see what it

561
00:31:51,908 --> 00:31:52,900
looks like.
Thanks.

562
00:31:53,000 --> 00:31:53,600
Okay.
Great.

563
00:31:53,600 --> 00:31:58,000
See ya.
That one was really good, pretty

564
00:31:58,000 --> 00:32:00,200
good.
That was getting loud better.

565
00:32:00,300 --> 00:32:03,000
Yeah, yeah.
So I'm just saying there's just

566
00:32:03,000 --> 00:32:06,200
all kinds of options here, you
know what, another thing I was

567
00:32:06,200 --> 00:32:09,400
just thinking about it's also
the pacing.

568
00:32:10,000 --> 00:32:13,600
It is the pacing to make it
sound because you don't want to

569
00:32:13,600 --> 00:32:17,200
talk like this because that's
not normal right, right?

570
00:32:17,200 --> 00:32:20,500
Right.
You automatically take pauses or

571
00:32:20,500 --> 00:32:22,500
you breathe.
And yeah.

572
00:32:22,600 --> 00:32:26,500
When I say Alicia, I might say
Alicia or hey Alicia.

573
00:32:26,500 --> 00:32:31,400
You have only yeah the tones
will change and like the Cadence

574
00:32:31,400 --> 00:32:34,200
of your conversations going to
change your your the words that

575
00:32:34,200 --> 00:32:38,400
you're saying yeah that I do see
that but yeah fuck that one of

576
00:32:38,400 --> 00:32:41,300
the one was pretty good.
Pretty good.

577
00:32:41,600 --> 00:32:47,300
Wow yeah that's fucking crazy.
So let's talk about how deep

578
00:32:47,300 --> 00:32:49,900
Lakes are created kind of the
it's explain the technology.

579
00:32:49,900 --> 00:32:55,100
So The technology behind deep
fakes is based on kind of

580
00:32:55,100 --> 00:32:57,300
mentioned earlier, learning
algorithms, which are a type of

581
00:32:57,300 --> 00:33:02,300
artificial intelligence that can
learn, and recognize and analyze

582
00:33:02,300 --> 00:33:07,300
complex patterns in data.
And while it may be hard to

583
00:33:07,300 --> 00:33:14,100
reconcile in your mind video,
Imagery and audio are nothing

584
00:33:14,100 --> 00:33:16,200
but patterns, right?
Yeah.

585
00:33:19,100 --> 00:33:20,900
Defects.
Use a type of deep learning

586
00:33:20,900 --> 00:33:27,500
algorithm called generative
adversarial, Network or gain.

587
00:33:28,500 --> 00:33:30,000
Now you go.
What the fuck is K?

588
00:33:30,000 --> 00:33:34,600
Yeah, I don't, I don't like the
adversarial part of it, I can

589
00:33:34,600 --> 00:33:36,900
system.
So this thing this is why it's

590
00:33:36,900 --> 00:33:42,100
adversarial, okay?
Gaines consists of two neural

591
00:33:42,100 --> 00:33:47,200
networks.
One is a generator Network.

592
00:33:48,600 --> 00:33:51,700
The other is a discriminator
network.

593
00:33:53,500 --> 00:33:57,000
Now, the generator network is
trained on a large data set of

594
00:33:57,000 --> 00:33:59,800
images or videos or sounds.
Whatever it is that you're

595
00:33:59,800 --> 00:34:03,900
trying to do.
And so, then it learns how to

596
00:34:03,900 --> 00:34:08,600
create those new images of the
videos that are similar to the

597
00:34:08,600 --> 00:34:12,600
original data set that it was
given sure the discriminator

598
00:34:12,600 --> 00:34:16,400
network on the other hand, it's
trained to differ if terrific,

599
00:34:16,400 --> 00:34:21,699
oh my God, differentiate between
real images and videos and those

600
00:34:21,699 --> 00:34:25,800
that are created by the
generator Network, okay?

601
00:34:25,800 --> 00:34:28,500
And this is somehow Works in
conjunction.

602
00:34:28,500 --> 00:34:32,500
The discriminator Works, to help
the generator because Knows the

603
00:34:32,500 --> 00:34:35,199
difference between real and
fake.

604
00:34:36,699 --> 00:34:38,199
Oh, okay.
Gotcha.

605
00:34:38,300 --> 00:34:41,199
Yeah, interesting, it is
interesting.

606
00:34:42,000 --> 00:34:44,699
And so as the generator Network
creates new images or videos,

607
00:34:44,699 --> 00:34:47,600
the discriminator network
evaluates them, and prevent

608
00:34:47,600 --> 00:34:49,800
provides feedback to the
generator.

609
00:34:51,000 --> 00:34:52,800
So it's helping the Buddy out.
Yeah.

610
00:34:52,800 --> 00:34:54,600
It's talk talking to each other.
Yep.

611
00:34:55,199 --> 00:34:58,100
So this feedback helps the
generator Network to refine its

612
00:34:58,100 --> 00:35:01,900
output and create more realistic
images and videos over.

613
00:35:02,000 --> 00:35:05,000
Time.
The generator becomes better and

614
00:35:05,000 --> 00:35:08,200
better better at creating all of
those things and then it becomes

615
00:35:08,200 --> 00:35:11,000
more difficult for the
discriminator network to

616
00:35:11,000 --> 00:35:16,200
differentiate the two, okay?
So that so when it stops having

617
00:35:16,200 --> 00:35:19,900
notes or when it has less notes
like, okay, that's pretty good.

618
00:35:20,500 --> 00:35:24,000
That's fucking crazy, crazy.
It's a learning.

619
00:35:24,000 --> 00:35:26,200
Yeah.
Yeah, these systems, these

620
00:35:26,200 --> 00:35:31,400
algorithms are learning is
really now to create a deep

621
00:35:31,400 --> 00:35:33,100
fake.
The Later network is trained on

622
00:35:33,100 --> 00:35:36,800
a specific individuals voice
face mannerisms.

623
00:35:37,000 --> 00:35:39,600
And then using this very large
data, set of images and videos

624
00:35:39,600 --> 00:35:42,700
of that person and then once the
generator network has learned

625
00:35:42,700 --> 00:35:46,400
these patterns, it can then
generate new images or videos

626
00:35:46,400 --> 00:35:49,800
that mimic the appearance, the
speech and the behavior and

627
00:35:49,800 --> 00:35:53,400
that's how you get those things
like we saw for Barack Now why

628
00:35:53,400 --> 00:35:58,400
the technology behind the fakes?
Is rapidly advancing.

629
00:35:58,400 --> 00:36:00,500
There are still limitations for
what it can do.

630
00:36:00,500 --> 00:36:04,000
So, for example, it can be
difficult to create realistic.

631
00:36:04,000 --> 00:36:06,900
Deep fakes with individuals
whose features and movements are

632
00:36:06,900 --> 00:36:12,200
very distinct and unique, okay,
okay.

633
00:36:12,200 --> 00:36:16,400
And while deep face can be much
more convincing with like short

634
00:36:16,400 --> 00:36:20,900
little Clips like Tom Cruise
going, hey girl versus you know,

635
00:36:21,000 --> 00:36:25,500
a 3 minute monologue, right.
They can because they can, then

636
00:36:25,500 --> 00:36:29,200
they start to pick can appear.
Or unnatural or unrealistic,

637
00:36:29,900 --> 00:36:31,600
even short ones than longer
versions.

638
00:36:31,600 --> 00:36:34,200
But the shirt ones are just
obviously way easier.

639
00:36:34,300 --> 00:36:37,200
Of course because you do is you
have your blessing, my vacation

640
00:36:37,200 --> 00:36:39,300
time to judge.
Yes, exactly.

641
00:36:39,900 --> 00:36:41,100
Yeah.
Wow.

642
00:36:41,200 --> 00:36:45,700
So a president going I've just
launched nuclear bombs is gonna

643
00:36:45,700 --> 00:36:47,900
be way more convincing than him
talking for an hour.

644
00:36:47,900 --> 00:36:51,000
Yeah I've just not launch
nuclear bombs and blah blah

645
00:36:51,000 --> 00:36:53,800
blah.
Okay so I have another question

646
00:36:54,000 --> 00:36:59,600
in regards to Just the so those
voices that we're listening to

647
00:36:59,600 --> 00:37:01,200
or just just a thought behind
that.

648
00:37:01,200 --> 00:37:03,300
Yeah so this lady what was her
name again?

649
00:37:03,300 --> 00:37:05,700
Oh yeah.
Linda I think was under Johnson.

650
00:37:05,800 --> 00:37:11,100
Yeah, so Linda her voice is more
similar to Gwyneth.

651
00:37:11,100 --> 00:37:12,900
Paltrow's right.
Which is where we're pulling all

652
00:37:12,900 --> 00:37:14,400
this, like free information
from.

653
00:37:14,500 --> 00:37:19,000
Uh-huh.
So if we had more sample, like a

654
00:37:19,000 --> 00:37:21,100
complete data, like our sample
size?

655
00:37:21,100 --> 00:37:24,200
Yes.
Yes, of varying voices.

656
00:37:24,200 --> 00:37:26,800
Yes, then it would become even
more.

657
00:37:26,900 --> 00:37:29,300
Intelligent.
This is just a 3-1, right?

658
00:37:29,300 --> 00:37:31,700
Exactly.
So I'm straight ahead hours and

659
00:37:31,700 --> 00:37:33,900
hours, and hours, and hours, and
hours, and hours, and hours of

660
00:37:33,900 --> 00:37:36,200
Obama, which we do.
Yeah.

661
00:37:37,600 --> 00:37:40,600
The system.
Now, what that generator Network

662
00:37:40,600 --> 00:37:43,600
and that other what we call it
the generator, many

663
00:37:44,200 --> 00:37:47,000
discriminator network was it?
Yeah, discriminator Mater.

664
00:37:47,000 --> 00:37:51,000
Yeah, those two can work it out
way better, because they've got

665
00:37:51,000 --> 00:37:54,200
a lot more data.
That's fucking crazy.

666
00:37:54,400 --> 00:38:00,300
I yeah, that's okay.
I have so many doesn't it.

667
00:38:00,300 --> 00:38:03,200
Insert to make your mind like
melt a little bit like thinking

668
00:38:03,200 --> 00:38:06,800
about.
It all go down, rabbit holes.

669
00:38:06,800 --> 00:38:13,200
I think I think also it You
know, we used to write like

670
00:38:13,200 --> 00:38:16,500
think, okay, people who are more
measured, who are, who seemed to

671
00:38:16,500 --> 00:38:21,500
be like, you know, more
controlled giving speeches or

672
00:38:21,500 --> 00:38:24,200
acting or whatever.
The case may be like, that was

673
00:38:24,200 --> 00:38:27,500
like, that was something that we
looked at and said, oh, I think

674
00:38:27,500 --> 00:38:30,300
that's a goal, right?
Like, that's something like,

675
00:38:30,300 --> 00:38:35,100
like, we have our politicians,
who are incredibly verbose, but

676
00:38:35,100 --> 00:38:40,700
they're very calm collected,
like, going into the And yes.

677
00:38:42,500 --> 00:38:47,000
That's easily taken advantage of
by AI systems.

678
00:38:47,000 --> 00:38:49,700
Apparently, even more, even more
easy, right?

679
00:38:49,800 --> 00:38:53,700
Because people who are more
expressive people who have more

680
00:38:53,700 --> 00:38:58,100
into Nations or, you know, the
algorithms have a hard time.

681
00:38:58,700 --> 00:39:01,800
Just that's that's where the art
comes in, right, like, right.

682
00:39:01,800 --> 00:39:04,000
Like when we're talking about
computers computers aren't

683
00:39:04,000 --> 00:39:07,000
Artful.
They are systematic.

684
00:39:07,000 --> 00:39:10,700
Yes.
So if you are more like this is

685
00:39:10,700 --> 00:39:13,500
what I'm going to talk about.
This is Next point, this is how

686
00:39:13,500 --> 00:39:16,700
I'm going to like, of course,
that's going to be easier to

687
00:39:16,700 --> 00:39:18,300
replicate.
Exactly.

688
00:39:18,300 --> 00:39:22,400
So I wonder what the future
might look like, because maybe

689
00:39:22,400 --> 00:39:25,200
everyone's going to be more
expressive because it's harder

690
00:39:25,200 --> 00:39:29,500
to fool a to copy.
Yes.

691
00:39:29,500 --> 00:39:32,200
Yeah.
Maybe maybe maybe that's the

692
00:39:32,200 --> 00:39:33,200
future.
Instead of having our

693
00:39:33,208 --> 00:39:37,100
politicians who are like, very
systematic, very more animated.

694
00:39:37,100 --> 00:39:38,100
Yeah.
Maybe they're going to become

695
00:39:38,100 --> 00:39:40,800
more animated because I don't
know.

696
00:39:40,800 --> 00:39:43,000
I don't either.
It's Just so interesting.

697
00:39:43,000 --> 00:39:47,600
It kind of brings up.
It might bring up a dilemma that

698
00:39:47,600 --> 00:39:49,800
these folks have.
Like, I don't want to be easily,

699
00:39:49,800 --> 00:39:52,700
replicate like replicated and
yeah, clearly, they're looking

700
00:39:52,700 --> 00:39:56,700
for ways to change it.
Governments are right ways to

701
00:39:56,700 --> 00:40:00,300
combat that.
Yes in order to you know, stuff

702
00:40:00,500 --> 00:40:06,600
world war, three or something,
you know, it's so scary.

703
00:40:06,600 --> 00:40:07,600
Yeah.
Because you're talking about

704
00:40:07,600 --> 00:40:11,100
like nuclear bombs, like that's
something that you could very

705
00:40:11,100 --> 00:40:14,300
easily be saying.
It's, it's just, it's insane

706
00:40:14,300 --> 00:40:17,000
Turtle today.
That's a careless things.

707
00:40:17,100 --> 00:40:18,600
Absolutely.
Yeah.

708
00:40:19,700 --> 00:40:23,000
My name's I've joked around and
let like said something stupid.

709
00:40:23,000 --> 00:40:25,000
Just as a joke but taken out of
context.

710
00:40:25,000 --> 00:40:29,300
Yes, right now you can just say,
make it people say it whenever.

711
00:40:29,400 --> 00:40:33,100
Oh, absolutely oof.
So some of the tools and

712
00:40:33,100 --> 00:40:36,000
software used to create defects.
I'm just gonna go through a few

713
00:40:36,000 --> 00:40:39,700
lists and this is such a, this
is just scratching the surface.

714
00:40:39,700 --> 00:40:43,000
There's something called Deep
face lab This is a popular

715
00:40:43,000 --> 00:40:44,900
software used for creating deep
bakes.

716
00:40:44,900 --> 00:40:49,700
It's open source.
And lets people swap faces

717
00:40:49,700 --> 00:40:52,400
bodies backgrounds and videos,
all kinds of stuff.

718
00:40:52,400 --> 00:40:53,100
Sure.
Yeah.

719
00:40:54,300 --> 00:40:57,000
Whole body are the face swap,
that's what we used.

720
00:40:57,000 --> 00:41:01,100
Yeah, for the Playboy for the
Playboy was face swap.

721
00:41:03,000 --> 00:41:05,900
As a film called fake app.
Another popular tool for deep

722
00:41:05,900 --> 00:41:09,000
fakes.
Deep dream.

723
00:41:09,700 --> 00:41:10,800
Oh God.
What's that sag?

724
00:41:11,000 --> 00:41:13,700
I know you guys like that's the
one episode in your like, oh

725
00:41:13,708 --> 00:41:16,300
sounds like dream of the fishes
and hey, yeah, yeah.

726
00:41:16,300 --> 00:41:19,300
Hey to her dream.
There's a lot of conversations

727
00:41:19,300 --> 00:41:23,000
with the word dream.
This is a gets a Google software

728
00:41:23,000 --> 00:41:25,500
that uses deep learning
algorithms to create, surreal,

729
00:41:25,500 --> 00:41:29,200
images, and videos.
And I don't know if you all have

730
00:41:29,200 --> 00:41:35,200
recently, saw the Google Kind of
open the doors to, I can't

731
00:41:35,200 --> 00:41:38,600
remember.
I think it was 60 minutes. and,

732
00:41:39,900 --> 00:41:42,600
You can tell it in ink with
words.

733
00:41:44,600 --> 00:41:48,200
Create me a flying pink.
Taxicab flying over New York and

734
00:41:48,200 --> 00:41:53,500
it will do it.
Okay, so we've actually used an

735
00:41:53,500 --> 00:42:00,000
AI to generate images on our
Instagram and for there you go.

736
00:42:00,200 --> 00:42:02,500
So how your McDonald's.
Okay.

737
00:42:02,500 --> 00:42:04,600
The McDonald's you had to do it.
I did.

738
00:42:04,600 --> 00:42:06,000
Yeah.
Create an image for you so I

739
00:42:06,000 --> 00:42:08,200
told us not copyrighted, right?
Right.

740
00:42:08,400 --> 00:42:12,900
It's exactly what you want.
Yes, well, I mean, okay, I would

741
00:42:12,900 --> 00:42:16,300
say that it probably would Like
having used it, it probably

742
00:42:16,300 --> 00:42:18,000
isn't.
So, I think like the script or

743
00:42:18,000 --> 00:42:21,300
the prompt that I gave it was
and it's through Discord, like,

744
00:42:21,300 --> 00:42:26,100
there's this Discord that you
can go onto and you just say,

745
00:42:26,100 --> 00:42:31,700
like, you do like a specific
script and then you say Ronald

746
00:42:31,700 --> 00:42:36,200
McDonald on A Monopoly board.
Gotcha, which is what I said.

747
00:42:36,200 --> 00:42:39,000
Yes, and it generated multiple
images.

748
00:42:39,000 --> 00:42:43,900
So if you go onto our on to our
Instagram, you will see the

749
00:42:43,900 --> 00:42:46,000
image.
That was created and it looks

750
00:42:46,000 --> 00:42:49,100
like it looks like the all the
writing was kind of in Russian

751
00:42:49,100 --> 00:42:52,100
which is fucking weird and I
couldn't get the Russian out of

752
00:42:52,100 --> 00:42:54,400
it.
I only said, probably for

753
00:42:54,400 --> 00:42:58,000
copyright reasons, right?
Could be could be but like, it

754
00:42:58,000 --> 00:43:00,500
didn't look.
It didn't look like, you know,

755
00:43:00,500 --> 00:43:03,800
an esoteric language or was
Junior Monopoly board?

756
00:43:03,900 --> 00:43:05,300
You knew it was a Monopoly
board.

757
00:43:05,300 --> 00:43:06,100
Yeah.
Because you had like the

758
00:43:06,100 --> 00:43:07,800
get-out-of-jail-free card.
Yeah.

759
00:43:07,800 --> 00:43:11,400
Like our corner like yeah.
Are like Ronald McDonald with

760
00:43:11,400 --> 00:43:14,000
lots of money or Ronald?
There was another one, Ron.

761
00:43:14,200 --> 00:43:19,500
Old McDonald with in an
interrogation and like the, the,

762
00:43:19,700 --> 00:43:22,000
and then you could have it do
different versions of that.

763
00:43:22,000 --> 00:43:23,700
Yeah.
And you could say, hey, I like

764
00:43:23,700 --> 00:43:26,100
this version because it gave you
multiple versions.

765
00:43:26,300 --> 00:43:31,100
You say, I like this version
kind of like ripped off of that.

766
00:43:31,100 --> 00:43:35,700
Yeah, exactly.
And it's insane how quickly it

767
00:43:35,700 --> 00:43:38,100
can generate and nothing.
May tell it to do anything.

768
00:43:38,400 --> 00:43:42,200
I wonder like what if I mean you
and I but like eventually it's

769
00:43:42,200 --> 00:43:44,000
going to get to the point where
it could.

770
00:43:44,200 --> 00:43:48,500
All right, some images that are
you and I like flying on

771
00:43:48,500 --> 00:43:50,600
unicorns or whatever.
Like I would love that picture.

772
00:43:50,600 --> 00:43:52,300
We could do it.
I mean, we could literally do

773
00:43:52,300 --> 00:43:55,900
that today because you give it a
picture of you and I and say put

774
00:43:55,900 --> 00:43:57,700
us on some unicorns.
I mean, we could do that.

775
00:43:58,000 --> 00:43:59,600
We should have it do that.
We should do that for this

776
00:43:59,600 --> 00:44:00,600
episode.
Yeah, exactly.

777
00:44:00,600 --> 00:44:04,500
Yeah.
Okay, Tanner flow is open source

778
00:44:04,500 --> 00:44:06,900
machine learning.
It's used to create deep fakes,

779
00:44:06,900 --> 00:44:09,300
provide a platform for building
and training, neural networks,

780
00:44:09,300 --> 00:44:12,000
that once it allowed for more
complex pie torch is another

781
00:44:12,000 --> 00:44:14,600
one.
Faceswap I talked about Out.

782
00:44:14,900 --> 00:44:16,400
Eva a see.
What was it?

783
00:44:17,900 --> 00:44:23,700
Avatara fi another one, it
allows people to actually

784
00:44:23,700 --> 00:44:26,800
animate images and videos with
these deep learning, so you can

785
00:44:26,800 --> 00:44:30,600
do have more animated versions
of things.

786
00:44:31,900 --> 00:44:36,700
So we're just kind of in the
grand scheme of things right on

787
00:44:36,700 --> 00:44:39,600
The Cutting Edge and so it's
just and the way technology

788
00:44:39,600 --> 00:44:43,400
moves in our world.
It's the rapid pace of

789
00:44:43,400 --> 00:44:45,500
improvement is just what you
were just saying.

790
00:44:45,800 --> 00:44:47,600
The Obama.
What happened on?

791
00:44:47,700 --> 00:44:49,300
19.
Exactly.

792
00:44:49,300 --> 00:44:52,100
Just looking at that.
Like what while worried that was

793
00:44:52,100 --> 00:44:53,400
four years ago?
Oh yeah.

794
00:44:53,400 --> 00:44:55,200
Now we're like oh yeah.
I know you can totally tell it's

795
00:44:55,200 --> 00:44:57,100
fake.
Totally, that's fucking crazy.

796
00:44:58,000 --> 00:45:01,900
So obviously there's ethical
implementation implications for

797
00:45:01,900 --> 00:45:06,400
creating Jim Fixx, right?
Defects can be used to deceive

798
00:45:06,400 --> 00:45:09,600
people by creating false or
misleading content that appears

799
00:45:09,600 --> 00:45:12,500
to be real.
And this can be harmful to

800
00:45:12,500 --> 00:45:15,600
people that make decisions based
on false information.

801
00:45:15,600 --> 00:45:18,700
Sure.
Where the defects used to defame

802
00:45:18,700 --> 00:45:23,400
or harm an individual they can
be used for deception and spread

803
00:45:23,400 --> 00:45:26,400
misinformation, which can then
can be, especially harmful if

804
00:45:26,400 --> 00:45:30,300
it's used to influence public
opinion or manipulate the

805
00:45:30,300 --> 00:45:34,700
outcome of an election or exact.
Absolutely defects can be used

806
00:45:34,700 --> 00:45:37,700
to violate an individual's
privacy by creating fake images

807
00:45:37,700 --> 00:45:39,800
or videos of them without their
consent.

808
00:45:40,500 --> 00:45:44,100
I'll basically you're
impersonating a person that can

809
00:45:44,100 --> 00:45:47,100
be damaging if the fake special,
especially if it's sexually

810
00:45:47,100 --> 00:45:49,100
explicit or defense.
Oh my God.

811
00:45:49,100 --> 00:45:52,400
Yeah.
And then people love to blame

812
00:45:52,400 --> 00:45:56,800
without facts or proof and it's
like guilty until proven

813
00:45:56,800 --> 00:45:59,700
innocent.
So yeah, yeah.

814
00:46:00,000 --> 00:46:02,000
People come out with their
pitchforks and I like, oh, you

815
00:46:02,000 --> 00:46:02,500
did.
Yes.

816
00:46:02,500 --> 00:46:03,200
And they're like no.
No.

817
00:46:03,200 --> 00:46:05,700
Actually I didn't and then you
have to.

818
00:46:05,700 --> 00:46:07,300
Yeah, you have to try to prove
yourself.

819
00:46:07,300 --> 00:46:12,600
Like I didn't, I didn't, Hose
for, you know, Playboy and I

820
00:46:12,600 --> 00:46:14,300
like ladies.
I'm looking at the picture right

821
00:46:14,300 --> 00:46:15,400
there.
Yes, you did.

822
00:46:15,500 --> 00:46:19,500
I was born in 87.
I was a fetus.

823
00:46:19,500 --> 00:46:21,500
Like, how do we know?
You know?

824
00:46:21,500 --> 00:46:23,700
And there's like no truth.
Then there's no.

825
00:46:24,400 --> 00:46:28,500
Then you can't, you can't have
like a there are no reliable

826
00:46:28,500 --> 00:46:30,800
sources.
Oh, right.

827
00:46:30,800 --> 00:46:34,400
I marry sources.
It's fucking scary, dude.

828
00:46:36,000 --> 00:46:38,900
Having a fake image audio file
or video of someone doing

829
00:46:38,900 --> 00:46:41,100
something illegal.
Oh my God.

830
00:46:41,100 --> 00:46:45,700
Yeah unscrupulous dangerous
morally unacceptable.

831
00:46:47,500 --> 00:46:52,700
I can never be undone, right?
And then lastly deep fakes can

832
00:46:52,700 --> 00:46:56,100
have legal Implement
implications as they can be used

833
00:46:56,100 --> 00:47:02,100
to create fake evidence or in
create or incriminate someone

834
00:47:02,100 --> 00:47:05,500
falsely.
Oh my God, that's true, that's

835
00:47:05,500 --> 00:47:08,300
true.
I didn't think of it in a right

836
00:47:08,300 --> 00:47:11,400
because you're not crazy, right?
Yeah.

837
00:47:11,400 --> 00:47:15,000
Oh yeah, there are plenty of
smart and crazy people.

838
00:47:15,100 --> 00:47:17,200
Yes, that's okay.
You should be that.

839
00:47:17,200 --> 00:47:20,600
Do your thing.
No, I'm not your enemy.

840
00:47:27,400 --> 00:47:30,800
So clearly lots of ethical
concerns and weeks, we really do

841
00:47:30,800 --> 00:47:36,700
need to start thinking through
the technology, but it's a very

842
00:47:36,700 --> 00:47:42,400
slippery slope because I mean,
it's important, it's clearly

843
00:47:42,400 --> 00:47:44,600
see, it's going to be important
to develop policies and

844
00:47:44,600 --> 00:47:48,900
regulations that can mitigate
the risks Associated to the Deep

845
00:47:48,900 --> 00:47:51,100
fakes.
But we have to ensure we are

846
00:47:51,700 --> 00:47:55,700
promoting Innovation and
creativeness in the field

847
00:47:55,700 --> 00:47:58,900
without censorship.
Right violating the first

848
00:47:58,900 --> 00:48:05,800
frickin Amendment, right, right?
There's an order to be creative,

849
00:48:06,000 --> 00:48:09,900
you have to kind of have to
allow yourself to have, as Many

850
00:48:09,900 --> 00:48:11,300
boundaries, right?
Right.

851
00:48:11,300 --> 00:48:19,400
So Where is that line?
That's where my brain starts to

852
00:48:19,400 --> 00:48:22,100
melt because I'm like, where is
that line?

853
00:48:22,300 --> 00:48:24,900
I think, I think traditionally
it's been ethically.

854
00:48:24,900 --> 00:48:26,200
Like right.
Okay.

855
00:48:26,200 --> 00:48:28,600
Doctors, things like that,
doctors and nurses.

856
00:48:28,600 --> 00:48:33,100
They take a vow of like, you
know, I don't know what it is.

857
00:48:33,100 --> 00:48:34,000
No huh?
Yeah.

858
00:48:34,000 --> 00:48:35,200
Yeah.
They're not gonna harm anyone.

859
00:48:35,200 --> 00:48:37,100
They're going to try to help
people.

860
00:48:37,300 --> 00:48:38,100
There's a space.
Yeah.

861
00:48:38,100 --> 00:48:39,200
There's a word for it.
Yeah.

862
00:48:39,200 --> 00:48:41,800
We just don't know.
Yeah, I know people are shouting

863
00:48:41,800 --> 00:48:45,300
and generate it right now.
Like like, hey, what have you

864
00:48:45,300 --> 00:48:48,500
not fucking know this not but
it's like, you know, you don't

865
00:48:48,500 --> 00:48:51,800
know.
Like, they take this vow, like,

866
00:48:51,800 --> 00:48:54,400
an ethical vow to say that
they're not going to do harm to

867
00:48:54,400 --> 00:48:57,900
someone that they're going to
try to help in any way possible.

868
00:48:57,900 --> 00:48:59,100
Yes.
No matter what, right?

869
00:48:59,100 --> 00:49:00,600
They're not going to use their
powers for bad.

870
00:49:00,600 --> 00:49:01,800
They're going to use it for
good.

871
00:49:02,000 --> 00:49:06,300
So what if there's some kind of
like ethical vow that we take

872
00:49:06,300 --> 00:49:09,900
because you can't one thing that
we've learned, at least, you

873
00:49:09,900 --> 00:49:12,300
know, both of us have learned on
this podcast, is that you can't

874
00:49:12,300 --> 00:49:14,700
fucking control people.
All right.

875
00:49:14,800 --> 00:49:19,900
You can't control them but you
know, having some kind of moral

876
00:49:19,900 --> 00:49:27,400
dilemma or ethical, like, you
know, agreement vow, then maybe

877
00:49:27,400 --> 00:49:31,900
perhaps that's a way to kind of
satiate these things but seems

878
00:49:32,000 --> 00:49:35,300
is sounds like a good idea but
as you just said, you can't

879
00:49:35,300 --> 00:49:37,900
control people and you certainly
can't fucking control

880
00:49:37,900 --> 00:49:41,300
technology.
No, you can't control these at

881
00:49:41,300 --> 00:49:42,500
least you.
And I can sit here and say,

882
00:49:42,500 --> 00:49:45,400
okay, we're going to take this.
Yeah, try totally are some way.

883
00:49:45,400 --> 00:49:47,000
I'm going to take the vow or
whatever, right.

884
00:49:47,000 --> 00:49:50,300
But like they don't have to what
about fucking programs have been

885
00:49:50,300 --> 00:49:55,100
programmed to do whatever and
like, you can't tell the program

886
00:49:55,500 --> 00:49:58,200
or the algorithm for the
ethical.

887
00:49:58,200 --> 00:50:00,100
It does not know what ethics
are.

888
00:50:00,200 --> 00:50:01,900
That's right, and that's what
they've done.

889
00:50:01,900 --> 00:50:03,900
A lot of these companies have
our already had to do.

890
00:50:03,900 --> 00:50:07,000
They've already learned their
lessons, like open API, they've

891
00:50:07,000 --> 00:50:10,000
had or open a, I think it's
opening.

892
00:50:10,000 --> 00:50:12,800
I, they've had to put blocks on
it because people have created

893
00:50:12,800 --> 00:50:15,600
some Easy effing shit.
Oh God.

894
00:50:15,700 --> 00:50:18,200
And so they've had to kind of
put some blockers in America

895
00:50:18,200 --> 00:50:22,200
because wall or something yes
some sort of a thing that helps

896
00:50:22,200 --> 00:50:24,900
the a I go.
No that's like morally not right

897
00:50:24,900 --> 00:50:29,000
order.
That's a little much like sensor

898
00:50:29,000 --> 00:50:31,200
doing is a little bit of
censorship obviously.

899
00:50:31,200 --> 00:50:35,000
Yeah.
Yes, see, that's the yeah, very

900
00:50:35,000 --> 00:50:36,300
slow.
That's exactly what you're

901
00:50:36,300 --> 00:50:38,200
talking about.
I yeah.

902
00:50:38,200 --> 00:50:39,800
What do you think about that?
Do you think?

903
00:50:39,800 --> 00:50:41,100
I don't know.
I don't know.

904
00:50:41,100 --> 00:50:43,100
That's why I said that's where
my mind starts to melt when I

905
00:50:43,100 --> 00:50:46,400
started thinking about this.
Because like, I just said, I

906
00:50:46,400 --> 00:50:54,300
think There's got to be some
boundaries, but don't want to

907
00:50:54,300 --> 00:50:57,700
step on people's Liberties.
Yeah, how do you create

908
00:50:57,700 --> 00:51:06,100
boundaries that are flexible.
So, there are some positive

909
00:51:06,100 --> 00:51:08,900
outcomes with this technology
entertainment.

910
00:51:09,100 --> 00:51:10,600
Sure.
It can be used to create new

911
00:51:10,600 --> 00:51:14,100
forms of entertainment videos
that combine faces different

912
00:51:14,100 --> 00:51:16,100
actors, or singers could be a
fun way to explore.

913
00:51:16,100 --> 00:51:18,300
New forms of media are
storytelling.

914
00:51:18,300 --> 00:51:21,700
Yeah, Education, create
educational content.

915
00:51:21,700 --> 00:51:26,100
That might, otherwise be
difficult to produce making and

916
00:51:26,100 --> 00:51:27,900
learning more interactive.
Hmm.

917
00:51:28,800 --> 00:51:34,000
Accessibility giving people more
access to content, such as

918
00:51:34,000 --> 00:51:37,000
videos with sign language.
For example, do have some build

919
00:51:37,000 --> 00:51:38,000
that for you.
Yeah.

920
00:51:38,800 --> 00:51:41,700
Research used to simulate
scenarios.

921
00:51:41,700 --> 00:51:44,600
That might not otherwise be able
to develop, but you can do that.

922
00:51:44,600 --> 00:51:49,000
Synthetically, special effects.
Yeah, Special.

923
00:51:49,100 --> 00:51:53,800
The movie industry.
So lots of good can come from

924
00:51:53,800 --> 00:51:54,500
it.
Sure.

925
00:51:56,300 --> 00:51:58,200
Just going to have to find a
balance.

926
00:52:00,100 --> 00:52:03,600
The kind of the skepticism and
regulation with the creative

927
00:52:03,600 --> 00:52:06,700
tivity in the growth and figure
out that about.

928
00:52:06,700 --> 00:52:07,800
I don't know what that's gonna
be.

929
00:52:08,900 --> 00:52:14,300
So how do they detect defects?
So there's a few ways there's

930
00:52:14,300 --> 00:52:18,200
visual inspection, you, and I
witnessed that metadata

931
00:52:18,200 --> 00:52:21,500
analysis, you can kind of get
the metadata of the thing to

932
00:52:21,500 --> 00:52:23,900
kind of better understand how it
was created, right?

933
00:52:24,400 --> 00:52:26,900
There's a thing called are Level
analysis.

934
00:52:27,500 --> 00:52:31,200
It involves analyzing an error.
Level of an image or a video

935
00:52:31,200 --> 00:52:34,000
that helps to identify if it's
been manipulated.

936
00:52:34,100 --> 00:52:36,400
Okay?
Interesting face recognition

937
00:52:36,400 --> 00:52:40,700
technology that the you can use
that audio analysis.

938
00:52:40,700 --> 00:52:44,300
I got of course making sure
things kind of match because as

939
00:52:44,300 --> 00:52:45,800
we heard sometimes you're a
little tinny.

940
00:52:45,800 --> 00:52:49,100
Yes, very tiny.
Yeah there's machine learning

941
00:52:49,100 --> 00:52:54,100
models that actually are build
being built To detect the things

942
00:52:54,100 --> 00:52:57,200
that the machine learning is
building, right, right, thank

943
00:52:57,200 --> 00:53:00,400
God.
It's cryptographic techniques

944
00:53:00,400 --> 00:53:02,800
like digital signatures and
watermarks that can I was going

945
00:53:02,800 --> 00:53:04,200
to say, yeah, I was going to ask
about that.

946
00:53:04,200 --> 00:53:05,600
I would assume that might be a
way.

947
00:53:05,608 --> 00:53:07,600
I think we do in the future.
Like if you're gonna publish

948
00:53:07,600 --> 00:53:12,300
something, got a water Market or
like those, what our QR codes or

949
00:53:12,300 --> 00:53:14,100
something like, you have a
different key are cut.

950
00:53:14,100 --> 00:53:16,900
I was thinking about that like,
yeah, in like a video, like,

951
00:53:17,000 --> 00:53:20,200
hey, this is actually us.
Here's our, here's our, we know,

952
00:53:20,200 --> 00:53:22,400
it's code for this week.
Yeah.

953
00:53:22,400 --> 00:53:25,500
Or for this, you know, whatever.
Exactly.

954
00:53:27,000 --> 00:53:29,000
But then you're probably
generating that through AI

955
00:53:29,000 --> 00:53:30,700
anyway.
So that's why it's like, you

956
00:53:30,700 --> 00:53:33,100
probably have to do it kind of
in the back end, right, where

957
00:53:33,100 --> 00:53:35,200
you don't really see it until
you've analyzed the video and

958
00:53:35,200 --> 00:53:38,700
then, you know, oh yeah, there's
back in Watermark that I yes,

959
00:53:38,700 --> 00:53:42,600
whatever.
Okay, yes, there's there's

960
00:53:42,600 --> 00:53:44,500
clearly no single method for
detecting.

961
00:53:44,500 --> 00:53:46,300
We're going to, it's going to
have to evolve.

962
00:53:46,300 --> 00:53:49,100
There's going to be many
different ways and what it is.

963
00:53:49,100 --> 00:53:53,800
It's expensive.
So, the spread of The Fakes,

964
00:53:53,800 --> 00:53:58,900
obviously complex multifaceted
approach, involving technology,

965
00:53:59,000 --> 00:54:01,100
technology and policy.
I thought, oh sure.

966
00:54:01,100 --> 00:54:06,400
Yeah.
So developing better detect and

967
00:54:06,400 --> 00:54:09,600
tection detection technology I
think is going to be important.

968
00:54:09,600 --> 00:54:13,600
I'm sure that's going to happen.
Yeah, promoting media literacy,

969
00:54:13,600 --> 00:54:16,700
like educating the public about
God, that's like sense of it.

970
00:54:16,700 --> 00:54:19,100
That's fucking huge.
The big one that is really huge.

971
00:54:19,100 --> 00:54:20,200
We still have that problem,
right?

972
00:54:20,200 --> 00:54:23,900
With like, do pigs like during
the last Allah, or what was it?

973
00:54:24,400 --> 00:54:28,500
The 2020 election.
Oh, yeah, sure is.

974
00:54:28,500 --> 00:54:30,200
There's like a lot of deep fake
super like oh no.

975
00:54:30,200 --> 00:54:31,600
Exactly.
Is definitely saying this?

976
00:54:31,600 --> 00:54:34,300
You're like, no.
No Trump definitely has a

977
00:54:34,300 --> 00:54:42,800
six-pack no definitely, no
definitely no, that's a big fat

978
00:54:42,800 --> 00:54:44,400
in.
Oh, did you have you seen some

979
00:54:44,400 --> 00:54:47,200
of those pictures?
Yes, I mean there's some of

980
00:54:47,200 --> 00:54:49,700
their so Preposterous, I mean,
doesn't look.

981
00:54:49,700 --> 00:54:52,800
Orange, is it super imposed like
do?

982
00:54:52,800 --> 00:54:56,800
I love the drama and the fun?
Yes, entertainment.

983
00:54:56,900 --> 00:54:59,900
But it's clear though.
No, but I, but people, but the

984
00:54:59,900 --> 00:55:04,400
ones that don't tell that's the
product that Is the probably

985
00:55:04,400 --> 00:55:06,600
used to be taught.
This is all right.

986
00:55:06,600 --> 00:55:08,700
You can use common sense are
going to talk about common

987
00:55:08,700 --> 00:55:10,400
sense.
I can I get on my soapbox?

988
00:55:11,000 --> 00:55:13,100
Okay, so we should have a little
segment.

989
00:55:13,100 --> 00:55:14,900
This is Alicia.
Soapbox.

990
00:55:15,400 --> 00:55:18,500
Get your cocktail in hand folks.
Wait, hold on.

991
00:55:18,500 --> 00:55:23,200
Wait, this is Alicia soapbox.
I'm ready.

992
00:55:24,000 --> 00:55:29,200
So ultimately, I feel like
oftentimes, we used to at least

993
00:55:29,200 --> 00:55:31,900
I was taught like, okay.
How can you tell what a source

994
00:55:31,900 --> 00:55:34,700
is?
Like, when I was in college,

995
00:55:34,700 --> 00:55:37,500
like, we would go through like
primary sources, things like

996
00:55:37,500 --> 00:55:39,600
that.
Even and Junior High, we weren't

997
00:55:39,600 --> 00:55:43,400
allowed to use Wikipedia ever,
right for articles.

998
00:55:43,400 --> 00:55:45,800
Like for research, we couldn't
use Wikipedia cause it wasn't

999
00:55:45,800 --> 00:55:47,100
legit.
Now it's like, oh my God,

1000
00:55:47,100 --> 00:55:48,700
actually, it's really fucking
legit.

1001
00:55:48,800 --> 00:55:51,100
It's an encyclopedia we used to
actually have to go through

1002
00:55:51,100 --> 00:55:53,900
encyclopedias my friends.
Oh yeah but we knew what a

1003
00:55:53,908 --> 00:55:57,700
primary source was we knew where
information was coming from and

1004
00:55:57,700 --> 00:56:01,500
whether or not that information
was sensationalized.

1005
00:56:01,500 --> 00:56:04,200
Whether or not there was someone
Trying to get us to think a

1006
00:56:04,208 --> 00:56:06,700
specific way.
Whether or not we were being

1007
00:56:06,700 --> 00:56:09,600
swayed.
We understood where the

1008
00:56:09,600 --> 00:56:12,000
conversation was was coming
from, right?

1009
00:56:12,100 --> 00:56:14,600
Yes.
So one thing that I thought was

1010
00:56:14,600 --> 00:56:17,300
really interesting is like,
during the last elections.

1011
00:56:17,300 --> 00:56:20,600
A lot of times people don't know
whether or not things that

1012
00:56:20,600 --> 00:56:22,600
actually been said by either
candidate, right?

1013
00:56:22,700 --> 00:56:26,600
Whether Democrat Republican
like, someone's like, oh I, I

1014
00:56:26,600 --> 00:56:29,700
want to kill babies like, nobody
wants to kill babies but like,

1015
00:56:30,000 --> 00:56:33,100
maybe there are things that are
pushed into that place.

1016
00:56:33,500 --> 00:56:36,400
Where they're saying, Obama
wants to kill babies and people

1017
00:56:36,400 --> 00:56:38,400
like, oh, of course, like I
said, we don't want to fucking

1018
00:56:38,400 --> 00:56:42,100
kill babies, but but, but people
believe that that's the case and

1019
00:56:42,100 --> 00:56:46,800
they believe that these deep
fake videos are reality instead

1020
00:56:46,800 --> 00:56:51,100
of sitting down and saying
exactly, he would never say

1021
00:56:51,100 --> 00:56:53,600
that, right?
Like this is clear someone, I

1022
00:56:53,600 --> 00:56:55,900
believe it.
Or do I really need to

1023
00:56:55,900 --> 00:56:58,000
fact-check?
Yes.

1024
00:56:58,000 --> 00:57:00,600
Do I need a fact Checker?
Like what is someone?

1025
00:57:00,600 --> 00:57:04,100
Trying Isabel pain?
Exact, first way my My opinion.

1026
00:57:04,100 --> 00:57:07,200
It's this, my friend, it's
totally that.

1027
00:57:07,700 --> 00:57:12,000
And I think that I think that
there is that has been lost.

1028
00:57:13,200 --> 00:57:17,600
And here, here, how?
I mean, I've seen up in be lost

1029
00:57:17,600 --> 00:57:19,800
like, I'm people who are a lot
older than me.

1030
00:57:20,000 --> 00:57:22,000
Yes.
And they believe these things

1031
00:57:22,000 --> 00:57:24,100
like, yes, it's true.
Common Sense.

1032
00:57:24,100 --> 00:57:27,300
Also like people are trying to
take advantage of you.

1033
00:57:27,400 --> 00:57:28,800
No matter what?
Right?

1034
00:57:28,800 --> 00:57:33,400
PT Barnum he said you know well
here's a said hey I suckers And

1035
00:57:33,400 --> 00:57:35,400
every minute, if whether or not
he did say that, whatever.

1036
00:57:35,400 --> 00:57:38,600
But like, but there are people
who are trying to sway you in a

1037
00:57:38,600 --> 00:57:42,600
specific direction for yes, for
there for a game for some kind

1038
00:57:42,600 --> 00:57:46,100
of game, weathers monetary or
whatever or your vote like, yes,

1039
00:57:46,100 --> 00:57:47,400
people are trying to take
advantage of.

1040
00:57:47,400 --> 00:57:50,900
Yes.
So how can I read into that?

1041
00:57:50,900 --> 00:57:53,800
Like if it seems like it's too
good to be true or like this is

1042
00:57:53,800 --> 00:57:56,400
the tail that I want to hear?
Yeah, it probably fucking is

1043
00:57:56,400 --> 00:57:59,100
yes.
So how do we how do we create

1044
00:57:59,100 --> 00:58:01,800
this media literacy?
How do we create Literacy for

1045
00:58:01,800 --> 00:58:03,500
these folks?
I think that's what we have to

1046
00:58:03,500 --> 00:58:09,200
figure out and other things like
strengthening content.

1047
00:58:09,200 --> 00:58:11,400
Moderation.
Again, I hate this.

1048
00:58:11,400 --> 00:58:14,300
The Priestly don't want to get
into censorship and all of that

1049
00:58:14,300 --> 00:58:21,100
shit but God damn it.
Like people can't use their

1050
00:58:21,100 --> 00:58:23,900
brains.
Well, they said like in our last

1051
00:58:23,900 --> 00:58:28,000
election or the election before
both of them, but the two

1052
00:58:28,000 --> 00:58:31,900
elections before like when it
was, let's see.

1053
00:58:33,200 --> 00:58:39,400
Umph and Obama.
Oh, a Biden Biden.

1054
00:58:39,700 --> 00:58:42,300
Was it, that was it?
Trump inviting before that the

1055
00:58:42,300 --> 00:58:46,900
second know who has Trump was
bhama, Obama.

1056
00:58:47,600 --> 00:58:48,900
Yeah.
And then it was Trump, and then

1057
00:58:48,900 --> 00:58:50,400
it was Biden.
Okay.

1058
00:58:50,500 --> 00:58:54,100
Well, yeah.
President's it like so, so when

1059
00:58:54,100 --> 00:58:58,700
Trump was wanting, oh, Hilary
Duff going from was running with

1060
00:58:58,700 --> 00:58:59,800
Hillary?
Yeah.

1061
00:58:59,800 --> 00:59:01,300
Okay.
So for the Hillary Trump

1062
00:59:01,300 --> 00:59:06,100
election, I'm with you.
They were saying that Russia.

1063
00:59:06,800 --> 00:59:09,400
They were having Bots or things.
How many?

1064
00:59:09,400 --> 00:59:12,100
Yes, and do.
All of the media posts?

1065
00:59:12,100 --> 00:59:15,000
Yes, they're trying to sway, the
American people.

1066
00:59:15,000 --> 00:59:15,600
Yes!
Yes.

1067
00:59:15,600 --> 00:59:19,000
Which is fucking scary, right?
Like, of course, we don't rock

1068
00:59:19,000 --> 00:59:23,800
rush out or fucking China, or
whomever to, to sway us, right?

1069
00:59:23,800 --> 00:59:27,200
But how do you create a media
literacy are people who are, how

1070
00:59:27,200 --> 00:59:29,900
do you do it?
How do you do it?

1071
00:59:30,000 --> 00:59:32,500
I don't know.
I don't know the answer clearly.

1072
00:59:32,500 --> 00:59:35,900
Like that's why we need the help
the right with moderation and

1073
00:59:35,900 --> 00:59:38,000
things like that.
Where you have moderate or any

1074
00:59:38,000 --> 00:59:39,000
companies.
Yeah.

1075
00:59:39,000 --> 00:59:40,400
Companies to help.
Yeah.

1076
00:59:40,800 --> 00:59:44,300
And as much as I hate to say it,
we're gonna need some

1077
00:59:44,300 --> 00:59:46,700
governmental help.
We need some oversight, sister,

1078
00:59:46,700 --> 00:59:50,800
how to set some level we do
need, we need oversight in

1079
00:59:50,800 --> 00:59:54,300
general.
Like because a lot of people

1080
00:59:54,800 --> 00:59:57,300
didn't grow up with with, with
technology, right?

1081
00:59:57,300 --> 01:00:00,000
Like, they don't even know how
to Email, right.

1082
01:00:00,000 --> 01:00:01,700
Let alone.
Look at a video of a private

1083
01:00:01,700 --> 01:00:03,600
video on YouTube.
They see a video.

1084
01:00:03,600 --> 01:00:05,200
It's true.
It's got to be real.

1085
01:00:05,200 --> 01:00:06,700
It has to be real.
Yeah.

1086
01:00:06,700 --> 01:00:09,700
Because it was on my Facebook
feed and Martha posted.

1087
01:00:09,700 --> 01:00:12,500
It's like zactly.
Martha is a fucking idiot.

1088
01:00:12,500 --> 01:00:14,000
Yeah.
She just repossessed something.

1089
01:00:14,000 --> 01:00:15,500
By the way, you just friended,
Martha.

1090
01:00:15,500 --> 01:00:17,500
Who is a robot, right?
Right.

1091
01:00:17,500 --> 01:00:19,300
Or people are falling.
Love with catfish and stuff.

1092
01:00:19,300 --> 01:00:22,000
That are not real people, right?
It's insane.

1093
01:00:22,100 --> 01:00:29,000
Exactly. so I mean overall
preventing the spread of deep

1094
01:00:29,000 --> 01:00:32,700
face is gonna like we said it's
going to require a coordinated

1095
01:00:32,700 --> 01:00:39,900
effort that involves technology
and, Oversight and policy.

1096
01:00:40,300 --> 01:00:44,200
Yeah, and I'd but I really think
by working together and using

1097
01:00:44,200 --> 01:00:48,600
Common Sense.
We can Ensure I think we can

1098
01:00:48,900 --> 01:00:50,700
help.
I don't think it will ever not

1099
01:00:50,700 --> 01:00:52,500
be a threat.
I think.

1100
01:00:52,600 --> 01:00:54,600
I think we will have some deep
fake problems.

1101
01:00:54,600 --> 01:00:57,900
Period period.
But I think we can help mitigate

1102
01:00:57,900 --> 01:01:03,800
it.
So what what do you think of

1103
01:01:03,800 --> 01:01:06,800
that does it make your brain?
Does this episode make your

1104
01:01:06,800 --> 01:01:09,500
brain process a little bit?
Yeah, I think.

1105
01:01:11,800 --> 01:01:19,400
Yes, it does.
I like to believe in the Star

1106
01:01:19,400 --> 01:01:25,400
Trek where this, okay?
I talked about the Star Trek

1107
01:01:25,400 --> 01:01:29,200
world because the Star Trek
world is amazing.

1108
01:01:29,800 --> 01:01:33,300
Yeah, tell them why.
Well, it's the belief in

1109
01:01:33,300 --> 01:01:37,400
humanity, it's the belief in
individuals that, you know, it

1110
01:01:37,400 --> 01:01:39,700
is an optimistic.
Yes, it's incredibly optimistic.

1111
01:01:39,700 --> 01:01:43,800
It's where you believe that
people are good and humanity is

1112
01:01:43,800 --> 01:01:47,300
worth saving human, you know,
humanoid whatever like,

1113
01:01:47,700 --> 01:01:53,000
life-forms life is important and
finding different life forms and

1114
01:01:53,000 --> 01:01:58,000
being curious about the world
only creates good.

1115
01:01:58,300 --> 01:02:01,300
And there's no.
So in the Star Trek universe,

1116
01:02:02,900 --> 01:02:06,400
oftentimes, you will have like,
my favorite is Star Trek next

1117
01:02:06,400 --> 01:02:11,000
Generation and this from and
Star Trek.

1118
01:02:11,000 --> 01:02:17,700
Next Generation is a scientific
vessel that goes just to explore

1119
01:02:17,700 --> 01:02:20,500
where no one has ever gone
before and to seek out new life

1120
01:02:20,500 --> 01:02:23,000
and new civilizations to boldly
go, where no one has gone

1121
01:02:23,000 --> 01:02:24,300
before.
That's the whole feel.

1122
01:02:25,200 --> 01:02:29,200
And so they're just going out of
curiosity and Trying to figure

1123
01:02:29,200 --> 01:02:31,700
things out and there is no
money.

1124
01:02:31,700 --> 01:02:33,400
There is no monetary gain,
right?

1125
01:02:33,400 --> 01:02:37,400
Money is no longer can you can
you believe it like everyone has

1126
01:02:37,400 --> 01:02:40,100
and how the world would change?
If money wasn't the driving

1127
01:02:40,100 --> 01:02:43,700
force behind something?
It would it's a different world.

1128
01:02:43,700 --> 01:02:50,200
It opens up the world to I'd
opens up Humanity to be able to

1129
01:02:50,200 --> 01:02:54,100
pursue what it is that you enjoy
and tend to be curious and seek

1130
01:02:54,100 --> 01:02:58,700
out knowledge or or create cool
things because Don't have to

1131
01:02:58,700 --> 01:03:01,600
worry about how you're going to
pay your bills or whatever, but

1132
01:03:01,600 --> 01:03:07,800
I believe that in general, we
want to do good and that

1133
01:03:07,800 --> 01:03:11,000
oftentimes, the reason why
people don't do good is because

1134
01:03:11,000 --> 01:03:15,200
they they have deeper seated
issues and they're hurting on

1135
01:03:15,200 --> 01:03:16,700
the Inside.
Yes.

1136
01:03:17,000 --> 01:03:20,700
Or they're trying to get money.
Yes.

1137
01:03:20,700 --> 01:03:22,500
Get money.
I mean just be simplistic about

1138
01:03:22,500 --> 01:03:24,700
it.
Yeah so if that wasn't a thing.

1139
01:03:24,700 --> 01:03:27,300
What's your motivation other
than just being an evil person,

1140
01:03:27,300 --> 01:03:29,100
right?
Yeah there are some Like, like

1141
01:03:29,400 --> 01:03:32,000
legitimately like evil people,
which is fine.

1142
01:03:32,000 --> 01:03:34,500
Yeah.
But I do believe, I believe

1143
01:03:34,500 --> 01:03:35,700
that.
Eventually.

1144
01:03:35,700 --> 01:03:38,100
We're going to try to we're
going to figure this out.

1145
01:03:38,200 --> 01:03:41,400
I hope so.
And and it's going to come to,

1146
01:03:41,500 --> 01:03:44,000
you know, things always get
worse before they get better.

1147
01:03:44,100 --> 01:03:45,200
Yeah, that's true.
That's true.

1148
01:03:45,300 --> 01:03:47,700
That's true.
So, we might just be at the Crux

1149
01:03:47,700 --> 01:03:51,500
or at the start of something
getting worse to where we have

1150
01:03:51,500 --> 01:03:53,700
to change things pretty and a
big bad way.

1151
01:03:53,700 --> 01:03:57,100
And I think II forces us it can
it can take care of some of

1152
01:03:57,100 --> 01:04:00,400
those like Medial tasks that,
you know, we don't want to have

1153
01:04:00,400 --> 01:04:06,000
to do like maybe data bullshit
typing in like 12 in an Excel

1154
01:04:06,000 --> 01:04:07,800
spreadsheet like maybe you don't
want to do that.

1155
01:04:07,800 --> 01:04:13,200
Maybe you could have a a I do
that but all right but but you

1156
01:04:13,200 --> 01:04:15,800
know and maybe people were
created to do something greater

1157
01:04:15,800 --> 01:04:18,100
and to have this curiosity,
maybe that's something that's

1158
01:04:18,100 --> 01:04:20,100
really important about Humanity
Star Trek.

1159
01:04:20,100 --> 01:04:23,400
There are there, are there.
Artificial intelligence are

1160
01:04:23,400 --> 01:04:25,800
always very very helpful.
They are.

1161
01:04:25,800 --> 01:04:31,500
Well depending oh, it's not.
Not not wait not so he's people,

1162
01:04:31,500 --> 01:04:37,700
hold on, not these people the
Borg.

1163
01:04:41,100 --> 01:04:48,000
The Borg is where AI inhumanity
merge and they you're watching

1164
01:04:48,000 --> 01:04:51,500
on video we are showing a borg
sphere.

1165
01:04:51,600 --> 01:04:54,200
Hmm.
Yeah they actually were cubes

1166
01:04:54,200 --> 01:04:58,000
initially that's the first thing
we discovered was the cubes then

1167
01:04:58,000 --> 01:05:00,800
they somehow evolved or maybe it
was a different quadrant and

1168
01:05:00,900 --> 01:05:04,000
spheres.
Yes, like we're not talking like

1169
01:05:04,000 --> 01:05:06,800
balls, we're talking planets.
Oh yeah.

1170
01:05:06,800 --> 01:05:09,000
It's like a planet.
It's like a whole ass planet.

1171
01:05:09,200 --> 01:05:12,100
Yeah.
And and ultimately, what they

1172
01:05:12,100 --> 01:05:16,800
did is they took the Borg would
take humanity and like everyone

1173
01:05:16,800 --> 01:05:20,000
was collected like in a hive
mind like Ai.

1174
01:05:20,000 --> 01:05:23,200
And so any any, it's a great
thing, right?

1175
01:05:23,200 --> 01:05:26,400
Like any kind of disease that
you had would be cured.

1176
01:05:26,400 --> 01:05:29,800
However, yeah.
You also didn't, you weren't an

1177
01:05:29,800 --> 01:05:33,000
independent mind, your own mind.
Yeah, everyone's mind is

1178
01:05:33,000 --> 01:05:37,400
together and it's all one and
when everyone is one, no one's

1179
01:05:37,400 --> 01:05:41,000
an individual is, right?
And they would assimilate.

1180
01:05:41,200 --> 01:05:45,700
So you wouldn't what resistance
is futile?

1181
01:05:46,000 --> 01:05:52,500
Hey say so So Alicia.
Yes.

1182
01:05:53,500 --> 01:05:57,000
Are you ready to have your mind
blown?

1183
01:05:57,000 --> 01:06:00,200
I ran out of your skull.
I already feel like I have so

1184
01:06:00,200 --> 01:06:02,200
many, like, moral dilemmas on
this one.

1185
01:06:04,300 --> 01:06:07,200
My mind's, a knots, and like all
over the place.

1186
01:06:07,200 --> 01:06:09,800
Getting all over.
Yeah, but yes, tell me here.

1187
01:06:09,800 --> 01:06:12,000
It goes, cab.
Hang on to your seat.

1188
01:06:13,700 --> 01:06:16,100
This entire episode was written
by an AI.

1189
01:06:19,200 --> 01:06:21,100
Shut up.
I'm not even kidding.

1190
01:06:21,100 --> 01:06:25,200
You know, are you serious?
I'm serious.

1191
01:06:25,300 --> 01:06:27,800
You had this episode written by
an AI.

1192
01:06:27,900 --> 01:06:30,100
Uh-huh.
No, yes.

1193
01:06:36,200 --> 01:06:39,400
I went to.
Yes, I went to open a, I and I

1194
01:06:39,400 --> 01:06:44,900
said, please write me a podcast
about deep fakes and it wrote

1195
01:06:44,900 --> 01:06:49,400
this episode.
And I asked, I asked for more

1196
01:06:49,400 --> 01:06:52,500
probing questions to get drill
down into stuff.

1197
01:06:52,700 --> 01:06:55,500
So, I had asked a total of five
questions and I had this whole

1198
01:06:55,500 --> 01:06:59,800
episode written ten minute how
many 10 minutes, how many pages

1199
01:06:59,800 --> 01:07:04,300
is it?
Oh it's Well, I mean I don't

1200
01:07:04,300 --> 01:07:08,800
know, I can scroll awhile well
but so maybe like five or six

1201
01:07:08,800 --> 01:07:11,100
pages.
Yeah, ten minutes.

1202
01:07:11,300 --> 01:07:12,100
You know?
How long?

1203
01:07:12,100 --> 01:07:14,400
A five or six?
Oh my God.

1204
01:07:14,900 --> 01:07:18,100
Podcast last me or takes me long
time?

1205
01:07:18,300 --> 01:07:20,300
It takes me.
Probably that.

1206
01:07:20,300 --> 01:07:22,300
What's going to do advertise?
Yeah.

1207
01:07:22,400 --> 01:07:25,300
A couple hours a day are yeah I
get it.

1208
01:07:26,800 --> 01:07:35,900
Are you fucking kidding me now?
Now listen, clearly I went and

1209
01:07:35,900 --> 01:07:38,200
made the little audio clips,
right?

1210
01:07:38,200 --> 01:07:40,000
I went made the way the audio
clips.

1211
01:07:40,300 --> 01:07:42,300
You.
You had to prompt it.

1212
01:07:42,400 --> 01:07:45,500
Yeah.
I went to another Ai and said,

1213
01:07:45,500 --> 01:07:47,500
you know, these are things.
I want you to say, and so, yeah.

1214
01:07:47,500 --> 01:07:55,000
So I had to go do all of that.
But I didn't, I mean, not much

1215
01:07:55,000 --> 01:07:57,500
more than that.
You didn't have to tell like, it

1216
01:07:57,500 --> 01:08:00,600
killed all of this out for me.
It put it all in bullet points.

1217
01:08:00,700 --> 01:08:05,200
It created it in segments.
It even said, like, can I see

1218
01:08:05,200 --> 01:08:11,900
that formatted at a little bit?
But yeah, so what a did here I

1219
01:08:11,908 --> 01:08:15,400
can actually show.
Yes, maybe it, can you see it?

1220
01:08:15,500 --> 01:08:19,500
So it says, like segment one.
You can see right here, says

1221
01:08:19,500 --> 01:08:22,500
Segment 1.
This is Deep fakes.

1222
01:08:23,000 --> 01:08:26,700
And then it started bullet izing
it out, and then giving me notes

1223
01:08:26,700 --> 01:08:28,399
and then it would do another
bullet.

1224
01:08:28,500 --> 01:08:30,600
And then I would say type of
deep fakes in.

1225
01:08:30,600 --> 01:08:34,600
It would ride it out and then I
would insert like my stuff like

1226
01:08:34,600 --> 01:08:37,500
the links T.
Okay, let's watch this little

1227
01:08:37,500 --> 01:08:41,899
episode, right?
And then we keep going down.

1228
01:08:42,899 --> 01:08:47,000
And then I there's a section of
this was my little section about

1229
01:08:47,000 --> 01:08:49,100
my little things, right?
And then it was segment two.

1230
01:08:50,700 --> 01:08:56,100
Oh my God and then little bullet
point of segments to I mean it

1231
01:08:56,100 --> 01:08:57,899
did all in the number did
everything.

1232
01:08:57,899 --> 01:09:00,600
It did it all.
That's fucking crazy.

1233
01:09:02,100 --> 01:09:05,200
That is insane to me.
Yeah I don't a computer wrote

1234
01:09:05,200 --> 01:09:09,700
this episode.
I don't know if I'm gonna sleep.

1235
01:09:09,700 --> 01:09:15,300
Okay like I really think I'm
gonna have some vicious like

1236
01:09:16,100 --> 01:09:19,899
that's got me fucked up I told
you you're mine was gonna be

1237
01:09:19,899 --> 01:09:23,000
blown.
I was talking to Dave about

1238
01:09:23,000 --> 01:09:24,800
this, huh?
And he's like you know you could

1239
01:09:24,800 --> 01:09:27,800
probably just like tell an AI to
like to write your episode

1240
01:09:27,800 --> 01:09:34,500
episode and I was like and I was
like I was like, no, you can

1241
01:09:34,500 --> 01:09:43,600
totally tell the difference.
Well, but here, but wait, you

1242
01:09:43,600 --> 01:09:46,600
totally?
Like we're creative like humans,

1243
01:09:46,600 --> 01:09:49,200
like all this shit.
Here you come.

1244
01:09:49,200 --> 01:09:52,200
I swear to God, we just had this
conversation like within the

1245
01:09:52,207 --> 01:09:55,500
past week crazy and you fucking
did it.

1246
01:09:56,500 --> 01:10:00,700
I mean, maybe I maybe I did.
I wonder I wonder so I don't

1247
01:10:00,700 --> 01:10:03,100
know, like, I couldn't I Don't
think I could have done this on,

1248
01:10:03,100 --> 01:10:05,800
like, the Texas Killing Fields.
I don't think I could have

1249
01:10:05,800 --> 01:10:08,700
gotten the detail that we got on
like McDonald's.

1250
01:10:08,700 --> 01:10:10,000
I don't think you could have got
the detail.

1251
01:10:10,000 --> 01:10:12,800
You did on that CI paper.
Yeah, I mean, you might have got

1252
01:10:12,800 --> 01:10:15,700
the CI paper, but I don't feel
they can get the background or

1253
01:10:15,700 --> 01:10:18,100
like, oh yeah, I just don't
question.

1254
01:10:19,200 --> 01:10:21,200
This one was very in a box,
right?

1255
01:10:21,200 --> 01:10:24,700
It was very kind of like,
systematic thematically blah,

1256
01:10:24,700 --> 01:10:28,200
blah, blah blah, where our other
episodes have been just kind of

1257
01:10:28,900 --> 01:10:31,400
more fluid.
I don't know how to describe it.

1258
01:10:31,400 --> 01:10:33,800
I just Don't I just don't think
it's there yet.

1259
01:10:33,900 --> 01:10:36,300
Like, I don't think it could
have written our other episodes.

1260
01:10:36,300 --> 01:10:37,800
Okay.
I don't.

1261
01:10:38,900 --> 01:10:41,800
It's worth trying.
Yeah I was just go see what it

1262
01:10:41,800 --> 01:10:43,100
would do.
Like go back to one of our

1263
01:10:43,100 --> 01:10:46,800
episodes and say tell them you
know, write a podcast about and

1264
01:10:46,800 --> 01:10:49,400
then fill in the blank.
What does it do with all this

1265
01:10:49,400 --> 01:10:53,600
information?
Like like I nerves it learned

1266
01:10:53,600 --> 01:10:58,900
that like earn but like I guess
where does this podcast go?

1267
01:11:00,000 --> 01:11:04,700
You mean the writing?
Yeah, the doesn't like store it.

1268
01:11:04,700 --> 01:11:07,800
You mean probably.
It probably does right to pull.

1269
01:11:07,800 --> 01:11:10,000
Maybe but part of me looks like
why?

1270
01:11:10,100 --> 01:11:13,400
Like why would it store it?
Like that wouldn't ya wouldn't

1271
01:11:13,400 --> 01:11:16,900
like what it probably Learned
was when I asked that and then I

1272
01:11:16,900 --> 01:11:24,200
asked and then after it / spit
out a good chunk, I asked it, I

1273
01:11:24,200 --> 01:11:29,200
said what software is used to
create deep fix.

1274
01:11:30,300 --> 01:11:32,000
And then remember it did that.
Yeah.

1275
01:11:32,000 --> 01:11:35,600
And then I said, what are the
ethical implications of defects

1276
01:11:36,400 --> 01:11:39,700
and then I said, what are the
methods of detecting deep fakes?

1277
01:11:40,000 --> 01:11:43,600
And then I said, what technology
can help control the spread of

1278
01:11:43,600 --> 01:11:47,200
deep fakes and that was
everything we just talked about

1279
01:11:47,500 --> 01:11:49,800
so what's in it?
So what it probably learned is,

1280
01:11:49,800 --> 01:11:53,000
well, these were the questions
that he asked relative to that

1281
01:11:53,000 --> 01:11:55,500
first question.
So it understands like what

1282
01:11:55,500 --> 01:11:58,000
content to put kind of put
together, right?

1283
01:11:58,000 --> 01:11:59,900
And so that it can help in the
future.

1284
01:12:00,300 --> 01:12:01,600
Create a better product.
Yes.

1285
01:12:01,600 --> 01:12:03,000
Yeah.
Like because it because it

1286
01:12:03,000 --> 01:12:05,800
starts out with that definition.
What I think is what I think is

1287
01:12:05,800 --> 01:12:12,200
so interesting is Yeah, it had
all of these details and but you

1288
01:12:12,200 --> 01:12:15,700
still had to prompt it to go
into the flow of.

1289
01:12:15,700 --> 01:12:18,900
I did, I did I didn't just say,
right the podcast and it spit

1290
01:12:18,900 --> 01:12:21,100
this out, right?
I had to do it in chunks.

1291
01:12:21,100 --> 01:12:23,700
Yeah, you had to kind of like
say okay, here's the flow of a

1292
01:12:23,700 --> 01:12:25,400
conversation.
Yeah, just like we're talking

1293
01:12:25,400 --> 01:12:30,300
about with like the intonation
or the Cadence of a person's

1294
01:12:30,300 --> 01:12:32,100
like, you know, voice our
conversation.

1295
01:12:32,100 --> 01:12:33,300
Like the way that they would say
something.

1296
01:12:33,300 --> 01:12:35,900
Yes, it doesn't know it
naturally, right?

1297
01:12:36,100 --> 01:12:38,100
You had to prompt that.
Yes.

1298
01:12:38,100 --> 01:12:40,800
And my delivery changes
everything for sure.

1299
01:12:40,800 --> 01:12:43,100
It does.
Yeah, you could just back and

1300
01:12:43,100 --> 01:12:45,700
says what it says exactly.
Yeah, had.

1301
01:12:45,700 --> 01:12:47,300
I read.
Yeah, exactly.

1302
01:12:47,300 --> 01:12:50,500
How I said, some of the tools
and software used to create

1303
01:12:50,500 --> 01:12:54,100
deep, fakes are deep face lab
face app.

1304
01:12:54,100 --> 01:12:55,700
Yes, right.
Yeah.

1305
01:12:55,700 --> 01:12:57,400
You're like, okay, that's so
robotic.

1306
01:12:57,400 --> 01:12:58,700
Yeah.
And it's not fun, it's not

1307
01:12:58,700 --> 01:13:01,800
interesting, right?
So they don't necessarily they

1308
01:13:01,800 --> 01:13:05,500
don't yet know.
Exactly what we're looking for,

1309
01:13:05,500 --> 01:13:07,400
but eventually I mean what
they're learning.

1310
01:13:07,400 --> 01:13:10,900
Yeah, they're learning, you
know, I don't know, 20, 25,

1311
01:13:11,600 --> 01:13:14,400
maybe I don't know everything,
maybe it, maybe you won't even

1312
01:13:14,400 --> 01:13:16,100
have to have to ask those
questions.

1313
01:13:16,100 --> 01:13:18,400
I know.
Fuck.

1314
01:13:19,600 --> 01:13:22,400
It's weird.
I don't know what I think I

1315
01:13:22,408 --> 01:13:24,900
don't either.
Does he now has having ethical

1316
01:13:24,900 --> 01:13:26,600
problems and that's exactly why
I wanted.

1317
01:13:26,700 --> 01:13:29,300
So, I did this for two reasons,
one, just to prove that it could

1318
01:13:29,300 --> 01:13:34,300
be done.
Yeah, but two, because we're

1319
01:13:34,300 --> 01:13:37,000
doing this episode.
I wanted to prove a point,

1320
01:13:37,000 --> 01:13:39,500
right?
I wanted to go like, we just did

1321
01:13:39,900 --> 01:13:42,300
what I asked Lee.
I was like, I don't like this.

1322
01:13:42,800 --> 01:13:46,400
Yeah, yeah, more like inter mice
morally.

1323
01:13:46,400 --> 01:13:48,700
I didn't like it, it feels.
It feels cheap.

1324
01:13:48,800 --> 01:13:51,400
Ichi T.
It feels shitty and authentic.

1325
01:13:51,400 --> 01:13:55,000
Yes, an authentic.
I feel like, it feels like kind

1326
01:13:55,000 --> 01:13:57,600
of dirty or something.
Yes, exactly.

1327
01:13:58,500 --> 01:14:02,000
Yeah, because we love to do this
and so we want, we want to put

1328
01:14:02,000 --> 01:14:04,900
effort into it, right?
And when you listen to do it for

1329
01:14:04,900 --> 01:14:06,900
me, no, no.
You listen to people like, you

1330
01:14:06,900 --> 01:14:09,100
know, doing a podcast, you're
like, oh I love like the bench,

1331
01:14:09,100 --> 01:14:12,100
they haven't things like that.
An AI is never going to be able

1332
01:14:12,100 --> 01:14:16,300
to create that bitch.
I mean not yet, but no, but I

1333
01:14:16,300 --> 01:14:18,700
mean the fact like you did it
perfectly.

1334
01:14:19,000 --> 01:14:20,900
We're on that.
We should have done, we are on

1335
01:14:20,900 --> 01:14:26,900
the fucking Edge and you so
beautifully showed us that like

1336
01:14:26,900 --> 01:14:32,100
that fucking my mind is blown.
I see, I don't know what to

1337
01:14:32,100 --> 01:14:34,900
think.
I don't know how to feel huh.

1338
01:14:34,900 --> 01:14:38,500
I feel like I'm going to be
thinking about this all night,

1339
01:14:38,600 --> 01:14:40,500
good.
I will probably be thinking

1340
01:14:40,500 --> 01:14:42,100
about this before I go to bed
tonight.

1341
01:14:42,600 --> 01:14:46,100
Like I feel, I hope, I hope so.
I hope I gave people something

1342
01:14:46,100 --> 01:14:48,000
to think about, we need to think
about.

1343
01:14:48,000 --> 01:14:49,200
We do need to think.
About it.

1344
01:14:49,200 --> 01:14:53,300
And I think I think, hey, if you
guys have any like, my father's,

1345
01:14:53,300 --> 01:14:54,800
our thoughts.
Yeah.

1346
01:14:54,800 --> 01:14:57,600
We have on our website and go to
the bazaar.

1347
01:14:57,600 --> 01:15:02,600
AF, just Spade, spelled out the
bizarre AF.com.

1348
01:15:02,800 --> 01:15:05,500
You can leave us a voicemail, we
can listen to it, or you can

1349
01:15:05,500 --> 01:15:07,900
send us an email info at the
bazaar.

1350
01:15:07,900 --> 01:15:11,400
AF.com, we had John air to like
we, yeah we can't.

1351
01:15:11,400 --> 01:15:12,800
What would play it and rock
crowd.

1352
01:15:12,800 --> 01:15:14,400
Answer your questions on you can
talk about it.

1353
01:15:14,400 --> 01:15:16,300
Yeah, I'd love to hear people's
thoughts on it.

1354
01:15:16,300 --> 01:15:18,600
Yeah, because this is a this is
a fucked.

1355
01:15:18,800 --> 01:15:23,800
This fucked up, fucked up.
One my God, thank you so much.

1356
01:15:23,800 --> 01:15:26,800
You're welcome for bringing this
moral dilemma that we should all

1357
01:15:26,800 --> 01:15:30,600
be thinking about front and
center.

1358
01:15:30,700 --> 01:15:33,100
Yeah, good, good, one.
That's thinking.

1359
01:15:34,000 --> 01:15:35,700
All right.
See everybody.

1360
01:15:35,800 --> 01:15:37,000
Yeah, baby.