1
00:00:00,240 --> 00:00:03,520
Speaker 1: Now here's a highlight from Coast to Coast AM on

2
00:00:03,640 --> 00:00:07,040
Speaker 1: iHeartRadio and welcome back to Coast to Coast George nor

3
00:00:07,280 --> 00:00:09,800
Speaker 1: with you along with John Wood, our special guest for

4
00:00:09,840 --> 00:00:13,039
Speaker 1: the next couple hours. John Wood is a partner in

5
00:00:13,080 --> 00:00:16,720
Speaker 1: the Green Client Trial Group. Sits on the advisory board

6
00:00:16,720 --> 00:00:20,599
Speaker 1: of the not for profit organizations dedicated to consumer protection

7
00:00:20,960 --> 00:00:25,040
Speaker 1: and sustainable business. John earned his Juris doctorate from New

8
00:00:25,120 --> 00:00:28,560
Speaker 1: York University School of Law. He is an author. The

9
00:00:28,600 --> 00:00:30,960
Speaker 1: book going to talk about tonight that he wrote with

10
00:00:31,000 --> 00:00:34,199
Speaker 1: the Native Sanders is called The Hugh Machine. It's a

11
00:00:34,320 --> 00:00:38,280
Speaker 1: very interesting book on artificial intelligence. His scholarship on human

12
00:00:38,360 --> 00:00:42,720
Speaker 1: nature and risk management have appeared in influential law review journals,

13
00:00:42,760 --> 00:00:46,400
Speaker 1: including NYUS Journal of Law and Liberty in the NYU

14
00:00:46,520 --> 00:00:50,919
Speaker 1: Environmental Law Journal. He has provided continuing legal education on

15
00:00:51,080 --> 00:00:55,240
Speaker 1: Internet law artificial intelligence to audiences including the New York

16
00:00:55,320 --> 00:00:59,720
Speaker 1: State Bar Association and Attorney General Alliance. Welcome John to

17
00:00:59,760 --> 00:01:02,920
Speaker 1: the program. Looking forward to chatting with you tonight. Thank

18
00:01:02,960 --> 00:01:05,480
Speaker 1: you so much, George. It's mutual great to be here.

19
00:01:05,959 --> 00:01:11,039
Speaker 1: The hu Machine A pronouncing that right, that's correct. Strange

20
00:01:11,120 --> 00:01:14,120
Speaker 1: thing about artificial intelligence. How did you as a lawyer

21
00:01:14,640 --> 00:01:19,120
Speaker 1: end up getting involved in writing about AI. Well, lawyers

22
00:01:19,200 --> 00:01:23,880
Speaker 1: are professional risk managers in a sense. And I've always

23
00:01:24,120 --> 00:01:27,760
Speaker 1: have been philosophically interested in things like philosophy of mind.

24
00:01:28,600 --> 00:01:32,840
Speaker 1: My mentor as a philosophy undergrad was a philosophy of

25
00:01:32,840 --> 00:01:37,080
Speaker 1: mind professor. And so I just care about risk. I

26
00:01:37,120 --> 00:01:39,560
Speaker 1: care about the risks that affect humanity. That's the kind

27
00:01:39,560 --> 00:01:42,400
Speaker 1: of thing that keeps me up at night. And this

28
00:01:42,480 --> 00:01:46,760
Speaker 1: is the biggest challenge facing humanity right now is coming

29
00:01:46,760 --> 00:01:50,800
Speaker 1: to grips with the artificial intelligence problem and the opportunity

30
00:01:50,840 --> 00:01:54,000
Speaker 1: associated with it. And when we talk about artificial intelligence,

31
00:01:54,080 --> 00:01:57,560
Speaker 1: John AI, what do we meet? What are we talking about? Well,

32
00:01:57,760 --> 00:02:00,560
Speaker 1: it's kind of a placeholder term that refers to, in

33
00:02:00,600 --> 00:02:05,720
Speaker 1: my mind, a constellation of different technologies that include computation,

34
00:02:06,160 --> 00:02:08,160
Speaker 1: which has been around for quite a while, but also

35
00:02:08,280 --> 00:02:13,440
Speaker 1: machine learning techniques and deep learning, neural networks, you know,

36
00:02:13,600 --> 00:02:17,720
Speaker 1: things like machine vision, any kind of any kind of

37
00:02:17,760 --> 00:02:22,640
Speaker 1: computer program that does computational tasks. We're also referring to

38
00:02:23,320 --> 00:02:27,280
Speaker 1: intelligent robots. So there's sort of the mental side of

39
00:02:27,280 --> 00:02:30,840
Speaker 1: things where the machines behaving like a human mind, and

40
00:02:30,880 --> 00:02:33,120
Speaker 1: then there's the physical side of things. Where the robots

41
00:02:33,120 --> 00:02:36,320
Speaker 1: are doing things that physical bodies used to do, and

42
00:02:36,360 --> 00:02:38,320
Speaker 1: we're referring to sort of all of those things, and

43
00:02:38,360 --> 00:02:40,320
Speaker 1: I'll try and be careful and delineate when I give

44
00:02:40,360 --> 00:02:43,639
Speaker 1: specific examples. Okay, Well, AI used to be really based

45
00:02:43,680 --> 00:02:46,680
Speaker 1: on science fiction. I mean, I remember movies based on

46
00:02:46,720 --> 00:02:49,880
Speaker 1: it was kind of fun and neat. It's kind of evolved,

47
00:02:49,919 --> 00:02:54,960
Speaker 1: testn't it. Absolutely? You know, it emerged as a serious

48
00:02:55,000 --> 00:02:57,240
Speaker 1: field of study in about the nineteen fifties, and they

49
00:02:57,320 --> 00:03:01,160
Speaker 1: predicted that we'll have general artificial intelligence in about twenty years,

50
00:03:01,520 --> 00:03:04,600
Speaker 1: and that timeline has kind of receded a day per

51
00:03:04,680 --> 00:03:07,440
Speaker 1: day every year since then. It's still about twenty years off,

52
00:03:07,440 --> 00:03:11,720
Speaker 1: it seems. But in the fiction and in movies, we

53
00:03:11,800 --> 00:03:14,520
Speaker 1: have a lot of very famous examples of AI. But

54
00:03:14,600 --> 00:03:17,400
Speaker 1: what's happened is that it's not fiction anymore. A lot

55
00:03:17,440 --> 00:03:21,519
Speaker 1: of publicly traded companies, a lot of technology startups are

56
00:03:21,600 --> 00:03:27,280
Speaker 1: rolling out robotic process automation, facial recognition technology, big data

57
00:03:27,360 --> 00:03:30,320
Speaker 1: driven machine learning algorithms, and a ton of new AI

58
00:03:30,440 --> 00:03:34,520
Speaker 1: based technology. It's in our inboxes, it's on social media,

59
00:03:34,840 --> 00:03:38,320
Speaker 1: our gps is are using it. So artificial intelligence has

60
00:03:38,360 --> 00:03:42,520
Speaker 1: really permeated our lives over the last ten years, it's

61
00:03:42,560 --> 00:03:47,040
Speaker 1: really taken hold. Can we say that artificial intelligence is

62
00:03:47,120 --> 00:03:52,640
Speaker 1: good for humanity and maybe bad for humanity? You know,

63
00:03:52,760 --> 00:03:57,200
Speaker 1: it's a very controversial topic. There is a very serious

64
00:03:57,240 --> 00:04:02,320
Speaker 1: debate happening within the philosophical ethical communities and technology ethics

65
00:04:02,320 --> 00:04:08,840
Speaker 1: and technology policy communities about the moral valance of these technologies.

66
00:04:09,080 --> 00:04:13,160
Speaker 1: I think there's some are by and large just bad.

67
00:04:13,200 --> 00:04:17,880
Speaker 1: There's a sort of universal consensus that facial recognition technology

68
00:04:17,960 --> 00:04:20,960
Speaker 1: is dangerous, for example, and that it is prone to

69
00:04:21,080 --> 00:04:25,479
Speaker 1: abuse and an invasion of privacy. But you know, it

70
00:04:25,560 --> 00:04:28,160
Speaker 1: kind of all depends on who's using it into what end.

71
00:04:29,080 --> 00:04:32,120
Speaker 1: You know, it's it's a complicated question. I mean, I

72
00:04:32,240 --> 00:04:38,800
Speaker 1: have been an opponent to driverless vehicles and driverless trucks. Well,

73
00:04:38,839 --> 00:04:40,920
Speaker 1: I don't think they're safe and I don't want to

74
00:04:40,920 --> 00:04:43,640
Speaker 1: see truck drivers put out of work. John, is that

75
00:04:43,720 --> 00:04:47,440
Speaker 1: an example of AI? It must be yes, that's a

76
00:04:47,880 --> 00:04:51,200
Speaker 1: great example of AI in our lives and basically teaching

77
00:04:51,240 --> 00:04:54,919
Speaker 1: a program to behave like a professional driver, how to

78
00:04:54,960 --> 00:04:59,160
Speaker 1: recognize the patterns and symbol of traffic and traffic signs,

79
00:04:59,640 --> 00:05:03,240
Speaker 1: and how navigate safely. You know, it's interesting I think

80
00:05:03,279 --> 00:05:05,560
Speaker 1: a lot of people are skeptical. No one wants to

81
00:05:05,600 --> 00:05:09,280
Speaker 1: see a traditional profession put out to work. We call

82
00:05:09,320 --> 00:05:12,240
Speaker 1: that bot sourcing, which is when instead of outsourcing to

83
00:05:12,360 --> 00:05:15,359
Speaker 1: other people somewhere else, you're outsourcing to a robot or

84
00:05:15,360 --> 00:05:18,920
Speaker 1: to a program. Now, and I think that we will

85
00:05:18,960 --> 00:05:25,920
Speaker 1: probably see relatively soon driverless cars that the perform is

86
00:05:25,960 --> 00:05:28,720
Speaker 1: safe or safer than human drivers, But I think we're

87
00:05:28,720 --> 00:05:30,880
Speaker 1: still going to be skeptical of adopting it in a

88
00:05:30,920 --> 00:05:35,080
Speaker 1: widespread manner. In nineteen sixty eight, space out of two

89
00:05:35,120 --> 00:05:37,680
Speaker 1: thousand and one came out, and you know, we all

90
00:05:37,760 --> 00:05:44,800
Speaker 1: remember how the computer we're doing its thing? Where are we? Well,

91
00:05:44,800 --> 00:05:50,320
Speaker 1: where are we now compared to that how computer? You know?

92
00:05:50,480 --> 00:05:54,359
Speaker 1: I think that we don't have anything like that now.

93
00:05:54,839 --> 00:06:00,600
Speaker 1: That's sort of generally intelligent control center of a spacecraft

94
00:06:00,720 --> 00:06:02,760
Speaker 1: or something that would be in charge of, say a corporation,

95
00:06:02,800 --> 00:06:05,560
Speaker 1: that's sort of an extended consciousness that runs the whole thing.

96
00:06:05,920 --> 00:06:09,000
Speaker 1: We don't have anything like that. The AI that we

97
00:06:09,080 --> 00:06:12,520
Speaker 1: have today is much more narrow. It can do sort

98
00:06:12,520 --> 00:06:17,320
Speaker 1: of extraordinary computational tasks with an extremely narrow confine, so

99
00:06:17,360 --> 00:06:19,880
Speaker 1: it can only really do what it's been programmed to do.

100
00:06:20,640 --> 00:06:24,000
Speaker 1: So the How from two thousand and one Space Odyssey

101
00:06:24,080 --> 00:06:27,560
Speaker 1: is still science fiction. We're not dealing with that kind

102
00:06:27,600 --> 00:06:30,000
Speaker 1: of intelligence. But I think that we should have a

103
00:06:30,040 --> 00:06:33,719
Speaker 1: serious discussion about what if we were to actually create

104
00:06:34,000 --> 00:06:37,440
Speaker 1: a general intelligence and it could become powerful and it

105
00:06:37,480 --> 00:06:40,240
Speaker 1: could do things like what how did and kill its

106
00:06:40,680 --> 00:06:44,760
Speaker 1: human users. But it's close. It feels like we're getting

107
00:06:45,080 --> 00:06:48,600
Speaker 1: very very close to that time period. It does, and

108
00:06:49,080 --> 00:06:51,359
Speaker 1: to hear the hype out of Silicon Valley, that is

109
00:06:51,400 --> 00:06:56,640
Speaker 1: their goal. That is nothing short of creating general artificial intelligence.

110
00:06:56,839 --> 00:06:59,440
Speaker 1: There's a brand new technology called deep fake. Kind of

111
00:06:59,480 --> 00:07:02,360
Speaker 1: explain what that is for us, Johnny. So. Deep bakes

112
00:07:02,480 --> 00:07:04,960
Speaker 1: is a form of synthetic media, and this is a

113
00:07:05,040 --> 00:07:09,360
Speaker 1: type of machine learning program. It's actually uses general adversarial networks.

114
00:07:09,360 --> 00:07:12,080
Speaker 1: To use a technical term, but basically, if I had

115
00:07:12,200 --> 00:07:16,600
Speaker 1: enough video footage of you talking, an audio feed of

116
00:07:16,680 --> 00:07:20,040
Speaker 1: you talking, I could feed that into this program and

117
00:07:20,080 --> 00:07:22,920
Speaker 1: then program in whatever type in whatever I want you

118
00:07:22,960 --> 00:07:25,880
Speaker 1: to say, and it could spit out what looks like

119
00:07:26,160 --> 00:07:29,280
Speaker 1: George talking and saying all the things I programmed in,

120
00:07:29,400 --> 00:07:32,120
Speaker 1: and it would be so close to life like that

121
00:07:32,400 --> 00:07:35,520
Speaker 1: your audience would have a very hard time telling the

122
00:07:35,560 --> 00:07:39,640
Speaker 1: fake apart from the real, authentic you. And that's why

123
00:07:39,720 --> 00:07:42,440
Speaker 1: we mean it's a deep fake. It takes very serious

124
00:07:42,480 --> 00:07:46,920
Speaker 1: forensic analysis to debunk a deep fake. And these are

125
00:07:47,000 --> 00:07:53,760
Speaker 1: potentially disastrous society, the dangerous. I mean, could you imagine

126
00:07:54,080 --> 00:07:56,760
Speaker 1: what you could do in a political campaign with something

127
00:07:56,840 --> 00:08:00,720
Speaker 1: like that. Absolutely, they can be weaponized, and you know,

128
00:08:00,840 --> 00:08:03,520
Speaker 1: you just need if they can actually play out in

129
00:08:03,560 --> 00:08:06,400
Speaker 1: two different ways. On one hand, I could create a

130
00:08:06,400 --> 00:08:09,080
Speaker 1: deep so a bad actor could create a deep fake

131
00:08:09,160 --> 00:08:12,560
Speaker 1: of their opponent or of a candidate that's got policy

132
00:08:12,800 --> 00:08:16,120
Speaker 1: views that hurt them in financially or something, and release

133
00:08:16,200 --> 00:08:19,000
Speaker 1: this deep fake to bring them down. On the other hand,

134
00:08:19,040 --> 00:08:21,040
Speaker 1: and a candidate could actually get caught in the act

135
00:08:21,160 --> 00:08:24,760
Speaker 1: doing something bad on camera and then say that wasn't me,

136
00:08:25,240 --> 00:08:27,440
Speaker 1: that was just a deep fake. Don't believe your eyes.

137
00:08:27,840 --> 00:08:31,520
Speaker 1: And that's what scholars call the liars dividend. That's interesting,

138
00:08:31,880 --> 00:08:34,440
Speaker 1: just the existence of deep fakes out there in the

139
00:08:34,520 --> 00:08:37,400
Speaker 1: ether means you can no longer believe what you're seeing.

140
00:08:37,480 --> 00:08:40,720
Speaker 1: It's something you have to fact check now everything you see. Basically,

141
00:08:40,760 --> 00:08:43,880
Speaker 1: they would do a reversal, They take what the real

142
00:08:43,920 --> 00:08:47,800
Speaker 1: truth was and then just deny exactly. Yeah, I'd say

143
00:08:47,800 --> 00:08:49,840
Speaker 1: that was a deep fake, and it's hard to debunk it.

144
00:08:49,880 --> 00:08:51,000
Speaker 1: I mean, what are you going to do issue a

145
00:08:51,040 --> 00:08:55,400
Speaker 1: press release? That's crazy. What happens one day if somebody

146
00:08:55,400 --> 00:08:58,800
Speaker 1: takes over some kind of cable or television network and

147
00:08:59,000 --> 00:09:02,040
Speaker 1: loads it up with all the deep fake stuff, well,

148
00:09:02,040 --> 00:09:05,040
Speaker 1: you know, it could certainly happen. And you know, at

149
00:09:05,040 --> 00:09:07,160
Speaker 1: the same time, like I said that the top, it

150
00:09:07,200 --> 00:09:09,560
Speaker 1: really just depends on how these technologies are used. So

151
00:09:09,679 --> 00:09:13,760
Speaker 1: you could imagine a president, you know, using deep fakes

152
00:09:13,760 --> 00:09:16,199
Speaker 1: as a public service announcement and saying, look, I might

153
00:09:16,240 --> 00:09:20,280
Speaker 1: not be available to issue this statement, but we can

154
00:09:20,360 --> 00:09:23,480
Speaker 1: use enough video footage of me to program in me

155
00:09:23,920 --> 00:09:26,160
Speaker 1: giving these series of statements, so I can travel the

156
00:09:26,280 --> 00:09:29,319
Speaker 1: country doing my executive duties and I can make announcements

157
00:09:29,320 --> 00:09:31,760
Speaker 1: at the same time. So there's ways to use deep

158
00:09:31,800 --> 00:09:35,240
Speaker 1: fakes in a good way for educational purposes, but obviously

159
00:09:35,320 --> 00:09:39,320
Speaker 1: they can use for deception, for corporate sabotage, for political influence.

160
00:09:40,240 --> 00:09:42,360
Speaker 1: So you know, there's a number of different state laws

161
00:09:42,360 --> 00:09:45,400
Speaker 1: that are cropping up, and even some federal legislation that's

162
00:09:45,440 --> 00:09:48,640
Speaker 1: cropping up, but in my analysis, they're all lacking they

163
00:09:48,640 --> 00:09:51,480
Speaker 1: have serious loopholes in them. John, is there money behind

164
00:09:51,559 --> 00:09:56,040
Speaker 1: AI right now? Absolutely? You can hear it on the

165
00:09:56,120 --> 00:09:59,320
Speaker 1: shareholder calls from big corporations, and you can see it

166
00:09:59,360 --> 00:10:02,160
Speaker 1: in the mass of valuations that tech companies are getting.

167
00:10:02,800 --> 00:10:06,320
Speaker 1: There's pretty much the next ten thousand startups. You could say,

168
00:10:06,480 --> 00:10:10,520
Speaker 1: just take some technology that exists now, add AI to it,

169
00:10:10,559 --> 00:10:13,079
Speaker 1: like yours tail, and boom, you've got a new product.

170
00:10:13,720 --> 00:10:16,120
Speaker 1: You know. I think that's sort of low hanging fruit

171
00:10:16,600 --> 00:10:21,160
Speaker 1: innovation with AI, and I'm calling for much deeper and

172
00:10:21,280 --> 00:10:24,320
Speaker 1: more meaningful applications of AI, which we can talk about.

173
00:10:24,800 --> 00:10:27,160
Speaker 1: But yeah, there's plenty of money in this space right now. Well,

174
00:10:27,160 --> 00:10:30,160
Speaker 1: with John Wood, we're talking about the Human Machine, which

175
00:10:30,240 --> 00:10:35,000
Speaker 1: is a really wordage for human machine, and he wrote

176
00:10:35,000 --> 00:10:38,480
Speaker 1: it with Nata Sandlers, who's a professor in Boston as well.

177
00:10:38,880 --> 00:10:44,240
Speaker 1: Is Nata scared with AI as well? Yeah? I think candidly.

178
00:10:44,920 --> 00:10:47,959
Speaker 1: You know, she's sent me an article the other day

179
00:10:48,040 --> 00:10:53,520
Speaker 1: about a company that's essentially destroying human privacy by scrubbing

180
00:10:53,559 --> 00:10:57,840
Speaker 1: the Internet and pulling everyone's faces and social media profiles

181
00:10:57,880 --> 00:10:59,840
Speaker 1: and sort of linking them all together. And this is

182
00:10:59,840 --> 00:11:03,560
Speaker 1: a technology that law enforcement has just bought on the

183
00:11:03,600 --> 00:11:06,040
Speaker 1: open market, and now they can you can basically take

184
00:11:06,080 --> 00:11:08,120
Speaker 1: a picture of someone's face and look up all sorts

185
00:11:08,160 --> 00:11:11,640
Speaker 1: of information about them. You know, a nightmare for stalking victims,

186
00:11:11,720 --> 00:11:15,880
Speaker 1: and you know, a dream come true for stalkers. And so, yeah,

187
00:11:15,960 --> 00:11:18,600
Speaker 1: she is concerned about the applications of those technologies, but

188
00:11:18,640 --> 00:11:22,600
Speaker 1: as a business professor, she has a strategic vision for

189
00:11:22,800 --> 00:11:26,320
Speaker 1: how corporate leaders can utilize AI to get the most

190
00:11:26,360 --> 00:11:28,679
Speaker 1: out of their human resource. I was going to say,

191
00:11:28,720 --> 00:11:30,960
Speaker 1: if you were a CEO, would you be tapping into

192
00:11:31,000 --> 00:11:34,400
Speaker 1: AI as much as you could? Right now? I would,

193
00:11:34,679 --> 00:11:37,520
Speaker 1: But then again, I have you know, I'm brave. Some

194
00:11:38,160 --> 00:11:40,360
Speaker 1: CEOs want to take a more cautious approach where they're

195
00:11:40,360 --> 00:11:43,000
Speaker 1: sort of on the sidelines, watching and seeing what competitors do.

196
00:11:43,559 --> 00:11:47,240
Speaker 1: There's a fair amount of companies that are experimenting with AI,

197
00:11:47,400 --> 00:11:50,360
Speaker 1: just sort of doing little pilot programs, and then the

198
00:11:50,440 --> 00:11:55,040
Speaker 1: minority of companies are actually diving in and using it

199
00:11:55,080 --> 00:11:59,480
Speaker 1: to transform their organization, their business strategy, and even their mission.

200
00:12:00,280 --> 00:12:02,960
Speaker 1: And so you know, there's this sort of bell curve,

201
00:12:03,040 --> 00:12:05,600
Speaker 1: bell shaped curve. So where the companies fall in terms

202
00:12:05,600 --> 00:12:08,400
Speaker 1: of how deeply are they integrating AI. Listen to more

203
00:12:08,520 --> 00:12:11,520
Speaker 1: Coast to Coast AM every weeknight at one a m.

204
00:12:11,600 --> 00:12:14,640
Speaker 1: Eastern and go to Coast to Coast am dot com

205
00:12:14,640 --> 00:12:15,040
Speaker 1: for more