1
00:00:00,240 --> 00:00:04,680
Speaker 1: From UFOs to psychic powers and government conspiracies. History is

2
00:00:04,760 --> 00:00:09,080
Speaker 1: riddled with unexplained events. You can turn back now or

3
00:00:09,200 --> 00:00:24,720
Speaker 1: learn the stuff they don't want you to know. M Hello,

4
00:00:24,760 --> 00:00:28,440
Speaker 1: and welcome back to the show. I'm no standing in

5
00:00:28,480 --> 00:00:31,160
Speaker 1: for Matt with that line, because whenever Matt's not here,

6
00:00:31,480 --> 00:00:32,959
Speaker 1: Ben and I Ben and I always look at each

7
00:00:32,960 --> 00:00:35,159
Speaker 1: other for a few minutes and realize that we're not

8
00:00:35,240 --> 00:00:38,920
Speaker 1: quite sure how to start the show. However, Never fear

9
00:00:38,960 --> 00:00:41,960
Speaker 1: Met is off on a very special secret project that

10
00:00:42,040 --> 00:00:44,640
Speaker 1: we cannot wait to tell you about. We will have

11
00:00:44,800 --> 00:00:47,479
Speaker 1: to tell you sometime soon, so look forward to that.

12
00:00:47,680 --> 00:00:50,360
Speaker 1: No spoilers. In the meantime, they call me Ben. We

13
00:00:50,479 --> 00:00:55,520
Speaker 1: are joined with our super producer, Paul the Personal Digital

14
00:00:55,520 --> 00:00:59,160
Speaker 1: Assistant decade. Yeah, p d A. But most importantly, you

15
00:00:59,240 --> 00:01:02,920
Speaker 1: are you. You are here, and that makes this stuff

16
00:01:03,040 --> 00:01:05,720
Speaker 1: they don't want you to know. We're we're pretty excited

17
00:01:05,720 --> 00:01:07,840
Speaker 1: about today's episode. What did p d A stand for

18
00:01:07,959 --> 00:01:10,600
Speaker 1: when it was like a pom pilot? Was that personal

19
00:01:10,600 --> 00:01:16,000
Speaker 1: digital Assistant? Uh? Yes, I believe so? Or was it okay?

20
00:01:16,000 --> 00:01:18,319
Speaker 1: So there was public display of Affection's that? But they

21
00:01:18,400 --> 00:01:22,440
Speaker 1: used to call like blackberries PDAs back before the advent

22
00:01:22,560 --> 00:01:26,520
Speaker 1: of the iPhone and the more tablet like device. Personal

23
00:01:26,560 --> 00:01:28,880
Speaker 1: digital assistant. That's what it has to be. Isn't that

24
00:01:28,880 --> 00:01:31,160
Speaker 1: funny though, because that's kind of become a new thing,

25
00:01:31,200 --> 00:01:34,199
Speaker 1: because today we're talking about the kinds of personal digital

26
00:01:34,280 --> 00:01:39,520
Speaker 1: assistance that you can talk to and can potentially talk back. Yes,

27
00:01:39,600 --> 00:01:41,640
Speaker 1: So let's look at some of the context. It's a

28
00:01:41,800 --> 00:01:46,800
Speaker 1: it's a very common trope in science fiction robots impersonating

29
00:01:46,920 --> 00:01:50,520
Speaker 1: human beings with increasing levels of fidelity, and we see

30
00:01:50,520 --> 00:01:52,680
Speaker 1: it in pop culture all the time, and some stories,

31
00:01:52,760 --> 00:01:56,560
Speaker 1: like in every episode of The Terminator, every everything related

32
00:01:56,600 --> 00:02:01,320
Speaker 1: to the Terminator franchise, machine consciousness tries to mimic humanity

33
00:02:01,360 --> 00:02:04,720
Speaker 1: exclusively as a means of waging war, and in other

34
00:02:04,760 --> 00:02:08,720
Speaker 1: places or other series, such as The Matrix, artificial intelligence

35
00:02:08,800 --> 00:02:12,240
Speaker 1: or AI attempts to surround us meat bags with an

36
00:02:12,240 --> 00:02:17,320
Speaker 1: impersonation of reality, complete with individual machine minds that can

37
00:02:17,360 --> 00:02:20,639
Speaker 1: pass for humans. And then in other cases, humans work

38
00:02:20,720 --> 00:02:24,560
Speaker 1: together to build technologies that can impersonate other human beings

39
00:02:24,720 --> 00:02:27,800
Speaker 1: in any number of ways, sometimes for sexy times reasons,

40
00:02:28,360 --> 00:02:31,760
Speaker 1: sometimes for just companionship. And we've seen that go awry.

41
00:02:31,800 --> 00:02:34,080
Speaker 1: I mean the Terminator, I think The whole point of

42
00:02:34,120 --> 00:02:37,880
Speaker 1: that series was that humans created Skynet or whatever for

43
00:02:37,919 --> 00:02:41,440
Speaker 1: their own to serve their own purposes, and then Skynet said,

44
00:02:42,200 --> 00:02:43,919
Speaker 1: what up, humans were done with? You are gonna do

45
00:02:43,960 --> 00:02:47,520
Speaker 1: our own thing or like in Westworld. And this goes,

46
00:02:48,240 --> 00:02:50,399
Speaker 1: like you said, in any number of ways, from increasingly

47
00:02:50,480 --> 00:02:55,880
Speaker 1: lifelike androids to entities that exist purely in the digital sphere,

48
00:02:55,919 --> 00:03:00,640
Speaker 1: able to hold genuine seeming conversations and fun shoeing at

49
00:03:00,760 --> 00:03:03,640
Speaker 1: or above what we would consider the average level of

50
00:03:03,720 --> 00:03:09,360
Speaker 1: human intelligence. As history has proven, science fiction is often prescient,

51
00:03:09,680 --> 00:03:13,520
Speaker 1: and it's not uncommon for authors to spin fantastic tales,

52
00:03:13,600 --> 00:03:18,560
Speaker 1: only for those tales to move years decades later from

53
00:03:18,600 --> 00:03:21,400
Speaker 1: the realm of science fiction to the world of science fact.

54
00:03:21,720 --> 00:03:25,079
Speaker 1: And the quest for this humanlike technology in real life

55
00:03:25,160 --> 00:03:28,640
Speaker 1: is no different. Yeah, I mean, I actually saw a

56
00:03:28,680 --> 00:03:31,919
Speaker 1: YouTube video a bunch of clips of times that science

57
00:03:31,960 --> 00:03:35,640
Speaker 1: fiction films have predicted technology that's totally a thing right now,

58
00:03:35,960 --> 00:03:39,040
Speaker 1: Like um Star Trek, for example, there's a little communicators

59
00:03:39,040 --> 00:03:43,080
Speaker 1: they're basically iPhones. And remember that scene in Total Recall

60
00:03:43,360 --> 00:03:45,640
Speaker 1: where they're walking through a full body scanner at the

61
00:03:45,680 --> 00:03:48,400
Speaker 1: airport and you see like their skeletons and you can

62
00:03:48,440 --> 00:03:50,680
Speaker 1: see the weapons hiding and stuff. That's pretty much what

63
00:03:50,720 --> 00:03:52,360
Speaker 1: we do now at the airport. We gotta put our

64
00:03:52,360 --> 00:03:55,000
Speaker 1: hands over our heads and stand in those full body

65
00:03:55,040 --> 00:03:57,680
Speaker 1: scanners where hopefully they just draw a little cartoon of

66
00:03:57,760 --> 00:04:00,480
Speaker 1: us with the naughty bits scrubbed out. But in theory

67
00:04:00,520 --> 00:04:04,480
Speaker 1: they can see everything. So too. Is AI and the

68
00:04:04,560 --> 00:04:07,000
Speaker 1: kinds of things we're talking about today. Yeah, and the

69
00:04:07,040 --> 00:04:11,480
Speaker 1: concept of what we call artificial intelligence. Long time listeners,

70
00:04:11,560 --> 00:04:14,680
Speaker 1: you may recall, as we use this phrase, you may

71
00:04:14,720 --> 00:04:18,120
Speaker 1: recall that some former guests of ours have objected to

72
00:04:18,240 --> 00:04:21,520
Speaker 1: the term artificial intelligence with a very good question, what

73
00:04:21,680 --> 00:04:24,880
Speaker 1: makes it artificial? What makes it if we're talking about consciousness,

74
00:04:24,880 --> 00:04:28,520
Speaker 1: what makes it any less of a consciousness than our own? Right?

75
00:04:29,000 --> 00:04:31,159
Speaker 1: And we think that's a good We think that's a

76
00:04:31,240 --> 00:04:35,560
Speaker 1: very good and valid point. Uh. We tend to agree

77
00:04:35,640 --> 00:04:38,159
Speaker 1: with it. But for the sake of brevity, we're just

78
00:04:38,200 --> 00:04:41,760
Speaker 1: gonna go with AI as a non pejorative thing. It's

79
00:04:41,800 --> 00:04:44,279
Speaker 1: just easier to say it that way. The concept of

80
00:04:44,279 --> 00:04:47,920
Speaker 1: AI has surprisingly old roots in our culture, especially if

81
00:04:47,920 --> 00:04:52,760
Speaker 1: we consider those ancient tales of non human entities impersonating

82
00:04:53,160 --> 00:04:57,200
Speaker 1: human beings. The ferry stories of changelings switched out at birth,

83
00:04:57,320 --> 00:05:00,880
Speaker 1: or or God's changing shape to breathe, eat with animals

84
00:05:01,000 --> 00:05:04,960
Speaker 1: or people, or shape shifters. In the twentieth century, the

85
00:05:05,000 --> 00:05:10,440
Speaker 1: concept of this artificial, inorganic thinking life form was popularized,

86
00:05:10,920 --> 00:05:13,039
Speaker 1: just like the point we made with science fiction through

87
00:05:13,080 --> 00:05:15,800
Speaker 1: culture and fiction, Like if you think of one of

88
00:05:15,839 --> 00:05:20,800
Speaker 1: the earliest artificial intelligences that blew up in the Western world,

89
00:05:21,120 --> 00:05:23,480
Speaker 1: it's the tin man in The Wizard of Oz. Yeah,

90
00:05:23,480 --> 00:05:24,960
Speaker 1: I never thought of him like that, but I guess

91
00:05:25,000 --> 00:05:26,960
Speaker 1: it's true. It's a good standing for that because even

92
00:05:26,960 --> 00:05:28,880
Speaker 1: the heart that he gets to the end spoiler alert

93
00:05:28,960 --> 00:05:31,760
Speaker 1: for the Wizard of Oz is like a clockwork cart,

94
00:05:31,960 --> 00:05:34,800
Speaker 1: you know, it's like something to add to his mechanations.

95
00:05:34,800 --> 00:05:37,400
Speaker 1: It's not actually a physical heart. So he's absolutely meant

96
00:05:37,440 --> 00:05:40,440
Speaker 1: to be like an automaton, which is kind of the

97
00:05:40,480 --> 00:05:43,640
Speaker 1: earliest form of robotics that go back to ancient times,

98
00:05:43,640 --> 00:05:48,640
Speaker 1: even where you have these incredibly intricate creations that move

99
00:05:48,839 --> 00:05:51,600
Speaker 1: through a series of gears and pulleys and what have you,

100
00:05:52,000 --> 00:05:54,279
Speaker 1: but are necessarily imbued with any kind of like ability

101
00:05:54,320 --> 00:05:56,000
Speaker 1: to make choices. But there are some that can even

102
00:05:56,040 --> 00:05:58,679
Speaker 1: like play games. I think there's one that's like a writer.

103
00:05:59,160 --> 00:06:04,480
Speaker 1: The mechanical term yeah, yeah, yeah, And they were lauded

104
00:06:04,520 --> 00:06:09,240
Speaker 1: for their ability to appear to do human esque things, right,

105
00:06:09,720 --> 00:06:12,800
Speaker 1: but people generally did not think they had a soul,

106
00:06:12,920 --> 00:06:14,880
Speaker 1: for instance. And it kind of goes back to you're

107
00:06:14,960 --> 00:06:17,520
Speaker 1: mentioning we had some folks say you had to take

108
00:06:17,600 --> 00:06:20,120
Speaker 1: issue with the term of artificial intelligence. We've also had

109
00:06:20,160 --> 00:06:22,039
Speaker 1: some folks right and taking issue with the term the

110
00:06:22,080 --> 00:06:26,240
Speaker 1: idea of machine learning, because you know, it is a

111
00:06:26,279 --> 00:06:29,240
Speaker 1: matter of we're still at a place where we have

112
00:06:29,320 --> 00:06:32,480
Speaker 1: to program machines to do what we want. We certainly

113
00:06:32,520 --> 00:06:35,160
Speaker 1: have have yet to fully experienced this like idea of

114
00:06:35,200 --> 00:06:38,919
Speaker 1: a singularity where the machine takes what we've imbuted it

115
00:06:39,000 --> 00:06:42,839
Speaker 1: with and develops its ability to go outside of that

116
00:06:42,960 --> 00:06:46,240
Speaker 1: or like make decisions outside of the parameters that we've programmed.

117
00:06:46,400 --> 00:06:48,200
Speaker 1: Very rarely have we seen that, and when we do

118
00:06:48,240 --> 00:06:52,440
Speaker 1: see it typically gets shut down. It's yeah, medic cognition, right,

119
00:06:52,720 --> 00:06:56,360
Speaker 1: so thinking about thinking rather so. By the nineteen fifties,

120
00:06:56,640 --> 00:07:01,040
Speaker 1: scientists and mathematicians and philosophers were for familiar with this

121
00:07:01,200 --> 00:07:05,960
Speaker 1: concept of quote unquote artificial intelligence, and our species began

122
00:07:06,040 --> 00:07:09,840
Speaker 1: to cognitively migrate from this world of fantasy toward a

123
00:07:09,880 --> 00:07:13,480
Speaker 1: world increasingly grounded in fact, and we can we can

124
00:07:13,520 --> 00:07:15,920
Speaker 1: hit some of the high points of artificial intelligence here.

125
00:07:16,120 --> 00:07:20,360
Speaker 1: In ninety the famous code breaker Alan Turing made one

126
00:07:20,400 --> 00:07:23,560
Speaker 1: of the most significant early steps in real life AI

127
00:07:23,640 --> 00:07:27,040
Speaker 1: when he and some of his colleagues created what we

128
00:07:27,120 --> 00:07:31,560
Speaker 1: now commonly call the Touring Tests. Named after him, of course,

129
00:07:31,720 --> 00:07:35,240
Speaker 1: he wrote this paper called Computing Machinery and Intelligence that

130
00:07:35,320 --> 00:07:39,560
Speaker 1: laid the groundwork both for the means of constructing AI

131
00:07:39,640 --> 00:07:42,840
Speaker 1: and for the ways in which we could measure the

132
00:07:42,880 --> 00:07:46,360
Speaker 1: intelligence of that AI or our success building it, which

133
00:07:46,440 --> 00:07:51,080
Speaker 1: might kind of be two different things. And sadly, well,

134
00:07:51,120 --> 00:07:54,640
Speaker 1: amazingly and sadly this line of thought was ahead of

135
00:07:54,640 --> 00:07:58,560
Speaker 1: the curve. Touring could not get right to work building

136
00:07:58,600 --> 00:08:03,040
Speaker 1: these human like minds were more specifically, he couldn't get

137
00:08:03,040 --> 00:08:06,680
Speaker 1: to work building minds that could full humans into thinking

138
00:08:06,880 --> 00:08:10,160
Speaker 1: they were also human minds because the technology at the

139
00:08:10,160 --> 00:08:12,920
Speaker 1: time had hard limits, like to the point you make

140
00:08:13,000 --> 00:08:17,800
Speaker 1: know about the mechanical turk. Up until nineteen nine or so,

141
00:08:18,240 --> 00:08:22,440
Speaker 1: computers couldn't really store commands we can say commands or

142
00:08:22,440 --> 00:08:25,720
Speaker 1: decisions here. They could only execute them and this meant

143
00:08:25,720 --> 00:08:28,880
Speaker 1: that the computers we could build at that time were

144
00:08:29,000 --> 00:08:33,199
Speaker 1: unable to satisfy a key prerequisite of intelligence. They couldn't

145
00:08:33,240 --> 00:08:37,400
Speaker 1: remember past events, past information, and therefore they could not

146
00:08:37,640 --> 00:08:42,080
Speaker 1: use this memory to inform present commands or decisions. And

147
00:08:42,240 --> 00:08:46,559
Speaker 1: in this way, computers began at that tabula rasas state

148
00:08:46,640 --> 00:08:50,320
Speaker 1: that blank slate state that so many mystic, spiritualist and

149
00:08:50,360 --> 00:08:57,320
Speaker 1: philosophers spend their lives attempting to attain I think that's fascinating. Computers,

150
00:08:58,080 --> 00:09:02,760
Speaker 1: like people who practice hardcore meditation, should existed mentally only

151
00:09:02,760 --> 00:09:05,079
Speaker 1: in the present. Yeah, And we'll get back to that

152
00:09:05,240 --> 00:09:07,800
Speaker 1: concept in a little bit in terms of um some

153
00:09:07,960 --> 00:09:11,160
Speaker 1: of the newer computers and how they are able to

154
00:09:11,400 --> 00:09:15,120
Speaker 1: quote unquote figure things out some better than others. But

155
00:09:15,280 --> 00:09:17,040
Speaker 1: let's just go back to the fifties for a second,

156
00:09:17,040 --> 00:09:21,040
Speaker 1: when computers were insanely expensive. I think this is no secret.

157
00:09:21,320 --> 00:09:23,800
Speaker 1: They ran you. You at least one, you couldn't even

158
00:09:23,840 --> 00:09:25,439
Speaker 1: own it. And that wasn't a thing for about two

159
00:09:25,840 --> 00:09:29,480
Speaker 1: K a month, which I believe been in today's standards.

160
00:09:29,559 --> 00:09:32,280
Speaker 1: That would be about two million dollars for one month

161
00:09:32,320 --> 00:09:36,600
Speaker 1: of computer use. In two dollars, a lot of Netflix subscriptions,

162
00:09:37,080 --> 00:09:40,959
Speaker 1: my friend um. And only of course the most prestigious

163
00:09:41,040 --> 00:09:46,439
Speaker 1: universities and huge technology corporations could even you know, afford

164
00:09:46,480 --> 00:09:49,800
Speaker 1: the cost of entry UM. So for Turing and his

165
00:09:49,920 --> 00:09:54,840
Speaker 1: ilk um to actually be able to succeed in building

166
00:09:55,040 --> 00:09:59,520
Speaker 1: something resembling anywhere closer artificial intelligence, they would have to

167
00:09:59,559 --> 00:10:04,280
Speaker 1: be part of some network of high profile, very wealthy

168
00:10:04,320 --> 00:10:09,800
Speaker 1: and influential UM funders of research. Yeah. That would that

169
00:10:09,880 --> 00:10:13,319
Speaker 1: they would be able to receive just an absolute boatload

170
00:10:13,320 --> 00:10:15,040
Speaker 1: of money from to get this kind of work done.

171
00:10:15,360 --> 00:10:17,720
Speaker 1: And imagine, you know, that's a hard sell. It seems

172
00:10:17,760 --> 00:10:22,000
Speaker 1: really interesting now, but back yeah, yeah, but back then

173
00:10:22,320 --> 00:10:24,920
Speaker 1: it was literally walking up to people and saying, hey,

174
00:10:24,920 --> 00:10:27,080
Speaker 1: we're good at math, and do you remember the tin

175
00:10:27,200 --> 00:10:29,920
Speaker 1: Man from Wizard of Oz. We sort of want to

176
00:10:29,960 --> 00:10:32,520
Speaker 1: make that. Can we have all of the money. So

177
00:10:32,559 --> 00:10:35,600
Speaker 1: that's that's tough, but they soldiered on. In ninety five,

178
00:10:35,640 --> 00:10:38,640
Speaker 1: another groundbreaking event occurred. There was the premiere of a

179
00:10:38,679 --> 00:10:41,920
Speaker 1: program called The Logic Theorists. It was supposed to mimic

180
00:10:41,960 --> 00:10:45,559
Speaker 1: the problem solving skills of a human being, and it

181
00:10:45,600 --> 00:10:50,000
Speaker 1: was funded by one of our favorite shady boogeyman, the

182
00:10:50,080 --> 00:10:54,360
Speaker 1: Research and Development corporation known today as Rand Rand Corporation.

183
00:10:54,440 --> 00:10:55,960
Speaker 1: That sounds like something out of a sci fi movie.

184
00:10:56,000 --> 00:10:58,520
Speaker 1: You don't know itself, And it's still around um doing

185
00:10:58,800 --> 00:11:01,600
Speaker 1: pretty interesting and secretive work because I believe they ended

186
00:11:01,679 --> 00:11:05,360
Speaker 1: up having a relationship with the US. Oh yeah, the

187
00:11:05,440 --> 00:11:10,960
Speaker 1: government big time. Absolute. You're absolutely right. This Logic Theorist

188
00:11:11,160 --> 00:11:15,720
Speaker 1: is considered by many people to be the first technically

189
00:11:15,760 --> 00:11:19,680
Speaker 1: the first artificial intelligence program. And there was a big

190
00:11:19,960 --> 00:11:24,200
Speaker 1: conference in nineteen fifty six hosted by John McCarthy and

191
00:11:24,240 --> 00:11:27,800
Speaker 1: a guy named Marvin Minsky, and in this conference they

192
00:11:27,880 --> 00:11:31,360
Speaker 1: presented the Logic Theorist as a proof of concept. The

193
00:11:31,360 --> 00:11:36,760
Speaker 1: conference is called the Dartmouth Summer Research Project on Artificial Intelligence.

194
00:11:36,760 --> 00:11:39,480
Speaker 1: And you know, no, I I thought of you with

195
00:11:39,520 --> 00:11:42,240
Speaker 1: this because I know how how we both love acronyms.

196
00:11:42,240 --> 00:11:43,680
Speaker 1: So I feel like in this one the D would

197
00:11:43,679 --> 00:11:46,240
Speaker 1: be silence. I'm just gonna call it surpi. That's great.

198
00:11:46,480 --> 00:11:48,880
Speaker 1: That sounds better because the D, you know, you can't

199
00:11:48,880 --> 00:11:56,760
Speaker 1: really do it. D s sound So the conference itself

200
00:11:56,800 --> 00:12:00,800
Speaker 1: fell short of the original very um bitious aims of

201
00:12:00,840 --> 00:12:03,520
Speaker 1: the organizers. They wanted to bring together the world's best

202
00:12:03,559 --> 00:12:06,560
Speaker 1: and brightest subject matter experts and by God, make an

203
00:12:06,640 --> 00:12:10,080
Speaker 1: artificial mind just in a week of a weekend, right, right,

204
00:12:10,480 --> 00:12:13,000
Speaker 1: they said it took God seven days. Let's uh, what

205
00:12:13,040 --> 00:12:14,599
Speaker 1: do you say? We could do it and for or

206
00:12:14,679 --> 00:12:18,640
Speaker 1: something like that. But the problem was pretty much everybody

207
00:12:18,679 --> 00:12:22,680
Speaker 1: disagreed on how exactly you would make a human like intelligence.

208
00:12:22,840 --> 00:12:25,200
Speaker 1: And at this point they're still thinking in terms of

209
00:12:25,679 --> 00:12:29,760
Speaker 1: artificial intelligence being like humans, which is a huge assumption.

210
00:12:30,040 --> 00:12:33,000
Speaker 1: But they unanimously agreed for the very first time on

211
00:12:33,040 --> 00:12:37,200
Speaker 1: a single crucial point that it was possible to make

212
00:12:37,440 --> 00:12:39,840
Speaker 1: a I and this set the stage for the next

213
00:12:39,920 --> 00:12:43,960
Speaker 1: two decades of research. So, from nineteen fifty seven to

214
00:12:44,120 --> 00:12:49,000
Speaker 1: nineteen seventy four, artificial intelligence interest in it or research,

215
00:12:49,080 --> 00:12:51,920
Speaker 1: and it really flourished. Computers they could stick, you know,

216
00:12:52,040 --> 00:12:55,080
Speaker 1: they were improving by leaps and bounds. They could store

217
00:12:55,120 --> 00:12:58,200
Speaker 1: more information, which is crucial, and of course they became

218
00:12:58,280 --> 00:13:02,280
Speaker 1: faster and cheaper and more excess double um. Machine learning

219
00:13:02,280 --> 00:13:05,000
Speaker 1: algorithms also began to improve, and people got better at

220
00:13:05,040 --> 00:13:09,240
Speaker 1: knowing which algorithm to use for their particular purposes, which

221
00:13:09,280 --> 00:13:11,520
Speaker 1: was also important because it wasn't you know that there

222
00:13:11,559 --> 00:13:15,240
Speaker 1: was sort of an established um language of algorithms that

223
00:13:15,240 --> 00:13:18,200
Speaker 1: people could pick and choose from to suit their particular problems.

224
00:13:18,760 --> 00:13:22,800
Speaker 1: Early successes like the general problem solver, Um, Ben, tell

225
00:13:22,840 --> 00:13:24,800
Speaker 1: me a little bit more about the general problem solver.

226
00:13:25,040 --> 00:13:27,240
Speaker 1: It does what it says on the tin. He could

227
00:13:27,320 --> 00:13:30,760
Speaker 1: give it a variety of problems and generally speaking, it

228
00:13:30,800 --> 00:13:33,000
Speaker 1: would attempt to solve them, not necessarily going to give

229
00:13:33,000 --> 00:13:36,200
Speaker 1: you the most creative solution. But yeah, and again that's

230
00:13:36,240 --> 00:13:39,199
Speaker 1: that's like an early days thing. So at the time

231
00:13:39,240 --> 00:13:43,640
Speaker 1: it was very impressive, but of course it was just

232
00:13:43,720 --> 00:13:47,280
Speaker 1: the beginning. And you know, you had another example I

233
00:13:47,360 --> 00:13:52,640
Speaker 1: think of uh Ai application in that time. Yeah, this

234
00:13:52,679 --> 00:13:56,760
Speaker 1: is when Alexa was first invented. Oh yes, yes, I'm kidding.

235
00:13:56,760 --> 00:14:00,320
Speaker 1: It's called Eliza. Um. But this is like a spoken

236
00:14:00,360 --> 00:14:03,120
Speaker 1: language interpreter which I don't know. I wonder if Alexa

237
00:14:03,280 --> 00:14:05,280
Speaker 1: is a is a nod to Eliza, what do you

238
00:14:05,280 --> 00:14:07,800
Speaker 1: think man? That would be pretty cool. Only a couple

239
00:14:07,800 --> 00:14:11,280
Speaker 1: of letters off. It was a spoken language interpreter um

240
00:14:11,320 --> 00:14:14,640
Speaker 1: that helped convince the government to really, Okay, this is

241
00:14:14,679 --> 00:14:16,880
Speaker 1: something that we're into and this is really important for

242
00:14:16,880 --> 00:14:22,440
Speaker 1: our story today. Yeah. It convinced the government, specifically DARPA

243
00:14:22,640 --> 00:14:29,000
Speaker 1: here in the United States, to start funding artificial Intelligence DARPA.

244
00:14:29,200 --> 00:14:32,160
Speaker 1: As you know if you are a longtime listener of

245
00:14:32,160 --> 00:14:36,760
Speaker 1: this show, DARPA is the resident mad science Department of

246
00:14:36,800 --> 00:14:41,960
Speaker 1: the United States government, and it stands for Defense Advanced

247
00:14:42,000 --> 00:14:46,200
Speaker 1: Research Projects Agency. They're the ones who do all the

248
00:14:46,440 --> 00:14:50,280
Speaker 1: X files level or fringe level stuff you hear about

249
00:14:50,320 --> 00:14:53,680
Speaker 1: in the news or often it's it's gonna be there

250
00:14:53,800 --> 00:14:56,680
Speaker 1: forefront of it. And then there's also stuff like Matt's

251
00:14:56,720 --> 00:15:00,640
Speaker 1: favorite public private partnership, skunk Works. It's a good name,

252
00:15:00,680 --> 00:15:03,000
Speaker 1: I mean, if you can look it for nothing other

253
00:15:03,040 --> 00:15:05,200
Speaker 1: than that, but they do all the secret Air Force

254
00:15:05,240 --> 00:15:09,120
Speaker 1: projects and you know, spy planes and all kinds of

255
00:15:09,120 --> 00:15:11,840
Speaker 1: stuff that's kind of technology you think about that is

256
00:15:11,880 --> 00:15:15,120
Speaker 1: probably years ahead of anything we've seen out there in

257
00:15:15,120 --> 00:15:19,520
Speaker 1: the world today, and it's staking for UFOs right uh

258
00:15:19,520 --> 00:15:23,880
Speaker 1: In this made people very gas very optimistic, because that's

259
00:15:23,920 --> 00:15:27,520
Speaker 1: another very human thing that we haven't quite learned how

260
00:15:27,600 --> 00:15:32,280
Speaker 1: to program optimism. And uh In nine seventy Marvin Minsky,

261
00:15:32,320 --> 00:15:35,480
Speaker 1: that guy who co hosted this conference, he told Life

262
00:15:35,480 --> 00:15:39,160
Speaker 1: magazine from three day years, we will have a machine

263
00:15:39,200 --> 00:15:42,640
Speaker 1: with the general intelligence of an average human being. He

264
00:15:42,800 --> 00:15:46,320
Speaker 1: was wrong, But to your point about suppressed technology, Noll

265
00:15:46,440 --> 00:15:49,880
Speaker 1: who's wrong so far as we know. It's interesting though

266
00:15:49,920 --> 00:15:52,160
Speaker 1: too not to get too far ahead of ourselves. But

267
00:15:52,280 --> 00:15:56,280
Speaker 1: with everything that's going on with big public facing companies

268
00:15:56,320 --> 00:16:00,280
Speaker 1: like Google and Amazon and Apple and stuff, you start

269
00:16:00,360 --> 00:16:03,200
Speaker 1: to get a sense that maybe the secret government stuff

270
00:16:03,240 --> 00:16:06,120
Speaker 1: isn't quite as far ahead as it once was, right,

271
00:16:06,280 --> 00:16:09,040
Speaker 1: I mean, that's just my opinion. I don't know. I

272
00:16:09,080 --> 00:16:11,240
Speaker 1: think it's a it's a good opinion. Something we want

273
00:16:11,280 --> 00:16:14,560
Speaker 1: to hear from you about, ladies and gentlemen, because it's

274
00:16:14,600 --> 00:16:20,280
Speaker 1: proven that in terms of physical hardware, material of war,

275
00:16:21,680 --> 00:16:25,200
Speaker 1: weapons and aviation and stuff, it is proven that the

276
00:16:25,320 --> 00:16:29,280
Speaker 1: US and most other governments want to keep that stuff

277
00:16:29,400 --> 00:16:32,080
Speaker 1: under wraps absolutely. I guess what I'm saying is when

278
00:16:32,080 --> 00:16:33,920
Speaker 1: you're when you're looking at a company that's trying to

279
00:16:34,000 --> 00:16:36,800
Speaker 1: sell you something, and you see how far they advance

280
00:16:36,880 --> 00:16:39,640
Speaker 1: with each update every year, you kind of get a

281
00:16:39,680 --> 00:16:42,400
Speaker 1: sense that maybe this is about where they're actually at.

282
00:16:42,480 --> 00:16:45,280
Speaker 1: That's that's the point. So I bring up the aviation

283
00:16:45,320 --> 00:16:48,680
Speaker 1: stuff and the weapons of war stuff to contrast it here,

284
00:16:48,800 --> 00:16:54,840
Speaker 1: because what what we're seeing also is that the governments

285
00:16:54,880 --> 00:16:58,320
Speaker 1: of the world. This is true, the governments of the

286
00:16:58,360 --> 00:17:02,280
Speaker 1: world usually can't pay the best and brightest as much

287
00:17:02,560 --> 00:17:06,639
Speaker 1: as the private entities game. So they're grabbing some of

288
00:17:06,680 --> 00:17:10,919
Speaker 1: the best workers unless those people are severely ideological, and

289
00:17:10,960 --> 00:17:14,520
Speaker 1: then they're making most of the progress which they would

290
00:17:14,600 --> 00:17:16,919
Speaker 1: later sell to a government. So I think maybe there

291
00:17:17,000 --> 00:17:19,240
Speaker 1: is a little more transparency. I think so, which is

292
00:17:19,359 --> 00:17:21,919
Speaker 1: which is an interestingly positive thing. But let let's le's

293
00:17:22,080 --> 00:17:24,400
Speaker 1: let's get keep, let's keep, let's keep huming right along.

294
00:17:24,680 --> 00:17:29,840
Speaker 1: So the quest for humanlike artificial intelligence soldiered on, but

295
00:17:29,880 --> 00:17:34,280
Speaker 1: it still had tremendous obstacles. Although computers could now store information,

296
00:17:34,560 --> 00:17:36,879
Speaker 1: they couldn't store enough of it, and they could not

297
00:17:37,080 --> 00:17:40,960
Speaker 1: process it at a fast enough pace. So funding dwindled

298
00:17:41,160 --> 00:17:45,920
Speaker 1: until the eighties. And that's when AI experienced a renaissance,

299
00:17:46,280 --> 00:17:49,200
Speaker 1: which will get to after a word from our sponsor.

300
00:17:55,080 --> 00:18:00,240
Speaker 1: So in the eighties, new tools and methods gave I

301
00:18:00,600 --> 00:18:04,200
Speaker 1: the field of AI a that renaissance that that Ben mentioned,

302
00:18:04,200 --> 00:18:07,520
Speaker 1: that kick in the digital pant and a guy named

303
00:18:07,600 --> 00:18:12,359
Speaker 1: John Hopfield and another fellow named David Rommelhart popularized this

304
00:18:12,440 --> 00:18:15,960
Speaker 1: concept of deep learning, which is a technique that allowed

305
00:18:15,960 --> 00:18:24,120
Speaker 1: computers to learn using experiences, using um paying attention to surroundings.

306
00:18:24,440 --> 00:18:27,520
Speaker 1: For example, the the idea of the fact that our

307
00:18:27,560 --> 00:18:30,760
Speaker 1: phones are serving us ads because we're talking about stuff,

308
00:18:30,760 --> 00:18:33,480
Speaker 1: and it's taking that information in and using it to

309
00:18:33,640 --> 00:18:37,040
Speaker 1: do something and learning our our habits, right, and this

310
00:18:37,080 --> 00:18:40,080
Speaker 1: is a really important part of things like these personal

311
00:18:40,280 --> 00:18:43,080
Speaker 1: digital assistance that we talked about at the top of

312
00:18:43,080 --> 00:18:47,119
Speaker 1: the show, right and on the other side of the

313
00:18:47,160 --> 00:18:49,720
Speaker 1: brain here too, or I guess. In a parallel approach,

314
00:18:50,240 --> 00:18:55,000
Speaker 1: a guy named Edward vegan Ball introduced expert systems that

315
00:18:55,200 --> 00:18:59,280
Speaker 1: mimicked the decision making process of a human expert. So

316
00:18:59,320 --> 00:19:03,320
Speaker 1: what happens. The program will ask an expert in the

317
00:19:03,359 --> 00:19:07,560
Speaker 1: field like it would ask super producer Paul or Richard

318
00:19:07,560 --> 00:19:11,159
Speaker 1: Feynman a question about production or physics, and then they

319
00:19:11,160 --> 00:19:13,920
Speaker 1: would say, how do you respond in this given situation?

320
00:19:14,240 --> 00:19:17,520
Speaker 1: And it could and it would take this for every

321
00:19:17,560 --> 00:19:20,520
Speaker 1: situation it could soak up, and then non experts could

322
00:19:20,600 --> 00:19:24,480
Speaker 1: later receive advice from that program about that information. It

323
00:19:24,520 --> 00:19:27,640
Speaker 1: sounds basic now. It sounds like how a search engine

324
00:19:27,880 --> 00:19:31,800
Speaker 1: can know, with increasingly accurate levels of fidelity, what question

325
00:19:31,880 --> 00:19:33,720
Speaker 1: you mean to ask when you ask you a question.

326
00:19:33,720 --> 00:19:36,720
Speaker 1: Remember when Google used to be super passive aggressive and

327
00:19:36,840 --> 00:19:40,240
Speaker 1: say did you mean? Well, now it just fills it

328
00:19:40,280 --> 00:19:43,000
Speaker 1: in for you Google instantly, and the reason for that

329
00:19:43,240 --> 00:19:46,760
Speaker 1: is that it has access to every entree that anyone

330
00:19:46,800 --> 00:19:49,399
Speaker 1: has ever put into Google, and so it combines all

331
00:19:49,400 --> 00:19:52,000
Speaker 1: that information and makes the best guess as to what

332
00:19:52,280 --> 00:19:55,400
Speaker 1: it thinks you're probably searching for based on what everyone

333
00:19:55,400 --> 00:19:57,440
Speaker 1: else that has started searching for that kind of thing

334
00:19:57,480 --> 00:19:59,879
Speaker 1: has also done, which is that same deep learning stuff

335
00:20:00,080 --> 00:20:03,200
Speaker 1: that is coming into play, which makes predictive text so

336
00:20:03,280 --> 00:20:06,720
Speaker 1: funny it can be. I guess it depends on the

337
00:20:07,080 --> 00:20:09,280
Speaker 1: platform you're on, and we'll get that a little bit too,

338
00:20:09,320 --> 00:20:12,440
Speaker 1: because sometimes it can be just really boneheaded. And haven't

339
00:20:12,440 --> 00:20:17,400
Speaker 1: you figured this out yet, Sirie spoiler alert, No he hasn't. Um.

340
00:20:17,440 --> 00:20:18,920
Speaker 1: But this was all really put to the test in

341
00:20:18,960 --> 00:20:22,320
Speaker 1: the nineties and two thousand's when we really hit some

342
00:20:22,480 --> 00:20:27,359
Speaker 1: big landmark moments and high water mark moments in artificial intelligence,

343
00:20:27,400 --> 00:20:30,160
Speaker 1: like with with Gary Kasprov who I know that you

344
00:20:30,280 --> 00:20:35,080
Speaker 1: are fascinated with, although he is a problematic figure at times. Yeah, yeah, Regie,

345
00:20:35,160 --> 00:20:40,400
Speaker 1: anti semitism aside. Yeah. Chess grand masters, the human ones

346
00:20:40,480 --> 00:20:47,560
Speaker 1: at least right are um immensely even fractally compelling, because

347
00:20:47,560 --> 00:20:50,919
Speaker 1: there's always some other layer to their personality, and you

348
00:20:51,000 --> 00:20:57,120
Speaker 1: have to wonder about the purported correlation between mental instability

349
00:20:57,280 --> 00:21:02,639
Speaker 1: and high thresholds of intelligence, because well, that's a story

350
00:21:02,680 --> 00:21:06,480
Speaker 1: for another day. Which that episode, Yeah, it might get

351
00:21:06,480 --> 00:21:09,960
Speaker 1: a little close down for us, but we can do it. So. Yeah,

352
00:21:10,000 --> 00:21:13,119
Speaker 1: as you're seeing, he was defeated by Deep Blue, built

353
00:21:13,119 --> 00:21:16,520
Speaker 1: by IBM, a computer just built to play chess. This

354
00:21:16,600 --> 00:21:20,000
Speaker 1: was a John Henry moment for the human race. And

355
00:21:20,040 --> 00:21:24,840
Speaker 1: then a scientist named Cynthia Brazel created Kismet, a program

356
00:21:24,880 --> 00:21:30,640
Speaker 1: of recognizing and displaying emotion. Now, for all our techno

357
00:21:30,760 --> 00:21:37,680
Speaker 1: futurist philosophers in the crowd, does Kismet recognizing and displaying

358
00:21:37,720 --> 00:21:43,439
Speaker 1: emotion automatically mean that Kismet experiences emotion? Story for another day.

359
00:21:43,480 --> 00:21:49,560
Speaker 1: But these uh, person versus program one on one matches

360
00:21:50,000 --> 00:21:53,680
Speaker 1: didn't stop with the game of chess. There was also Jeopardy, right,

361
00:21:54,320 --> 00:21:58,320
Speaker 1: And then there was one other one that had tremendous

362
00:21:58,320 --> 00:22:02,040
Speaker 1: implications for the world of programming. Yeah, and that was

363
00:22:02,119 --> 00:22:05,840
Speaker 1: much more recently with Google's Alpha go. Um let's go

364
00:22:06,080 --> 00:22:10,400
Speaker 1: as in the ancient Chinese strategy game of Go, and

365
00:22:10,600 --> 00:22:16,200
Speaker 1: it successfully defeated a Go champion Kaiji, and Go is

366
00:22:16,200 --> 00:22:22,360
Speaker 1: is a notoriously complex game, very difficult to predict, and

367
00:22:22,359 --> 00:22:24,680
Speaker 1: and always have to be many many moves ahead of

368
00:22:24,720 --> 00:22:28,400
Speaker 1: your opponent. So that's pretty cool. That's almost a step further.

369
00:22:28,400 --> 00:22:32,280
Speaker 1: Wouldn't you say, Ben, that Go would be more challenging

370
00:22:32,400 --> 00:22:36,399
Speaker 1: for a computer to beat a human opponent than chess?

371
00:22:36,440 --> 00:22:40,119
Speaker 1: Even like this is like a real serious leap forward.

372
00:22:40,359 --> 00:22:44,440
Speaker 1: I would absolutely agree, And a lot of people were

373
00:22:44,920 --> 00:22:49,720
Speaker 1: even more skeptical watching that game. People understood Go. We

374
00:22:49,760 --> 00:22:51,680
Speaker 1: can also go out there and say, Paul, I don't

375
00:22:51,680 --> 00:22:54,520
Speaker 1: know if you're familiar with this, but are you Are

376
00:22:54,560 --> 00:22:59,720
Speaker 1: you a Go enthusiast? You play this game? No? Not really, Noel,

377
00:23:00,000 --> 00:23:04,440
Speaker 1: I've never played it. I remember it from the movie Pie, Remember, Yeah,

378
00:23:04,920 --> 00:23:08,200
Speaker 1: the one, one of Darren Aronowski's first movies, which people

379
00:23:08,280 --> 00:23:10,240
Speaker 1: kind of crap on. But I enjoyed it a lot

380
00:23:10,280 --> 00:23:12,600
Speaker 1: when I was young anyway. But that movie is about

381
00:23:12,840 --> 00:23:18,560
Speaker 1: a kind of a mad computer scientist who realizes that

382
00:23:18,680 --> 00:23:23,560
Speaker 1: there that the Kabbala contains some kind of code involving

383
00:23:23,720 --> 00:23:27,240
Speaker 1: the number for pie and gets goes down a crazy

384
00:23:27,320 --> 00:23:30,159
Speaker 1: rabbit hole of insanity and paranoia. But Go as a

385
00:23:30,200 --> 00:23:34,720
Speaker 1: recurring game in that movie, implying that it's all about

386
00:23:35,040 --> 00:23:37,880
Speaker 1: high level, very very high level thinking. Yeah, I think

387
00:23:37,920 --> 00:23:41,199
Speaker 1: that's I think that's excellent. You know, Pie is a

388
00:23:41,200 --> 00:23:45,639
Speaker 1: fantastic but jarring film, you know. And so we see

389
00:23:46,119 --> 00:23:50,040
Speaker 1: a common trend in these again, we'll call them John

390
00:23:50,080 --> 00:23:53,760
Speaker 1: Henry moments. And people generally know the story of John

391
00:23:53,800 --> 00:23:55,680
Speaker 1: Henry right now, you think they do? I know he

392
00:23:55,800 --> 00:23:58,560
Speaker 1: was a steel driving man. Give me some more, well,

393
00:23:58,640 --> 00:24:02,000
Speaker 1: John Henry, anyone who was outside of the US. And

394
00:24:02,119 --> 00:24:04,439
Speaker 1: I don't think we we know for sure how common

395
00:24:04,440 --> 00:24:10,119
Speaker 1: this story is in the States anymore. He was in folklore.

396
00:24:10,600 --> 00:24:15,240
Speaker 1: He was African American steel driving man. Esneal said, that's

397
00:24:15,240 --> 00:24:18,440
Speaker 1: absolutely true. And his job was the hammer a steel

398
00:24:18,840 --> 00:24:22,159
Speaker 1: drill into rock to make holes for explosives, right, and

399
00:24:22,200 --> 00:24:26,840
Speaker 1: they would blast rock constructing a railroad tunnel. And in

400
00:24:26,880 --> 00:24:32,640
Speaker 1: the legend he was pitted against a new fangled steam

401
00:24:32,720 --> 00:24:37,359
Speaker 1: powered hammer. It was a man against machine race. And

402
00:24:38,520 --> 00:24:42,080
Speaker 1: the historical aspect of whether this did or didn't happen

403
00:24:42,320 --> 00:24:45,119
Speaker 1: is that people go back and forth on it. But

404
00:24:45,280 --> 00:24:47,919
Speaker 1: what we're experiencing now with AI is a series of

405
00:24:47,960 --> 00:24:53,760
Speaker 1: increasingly high stakes John Henry moments. Key g said, Gary

406
00:24:53,800 --> 00:24:58,920
Speaker 1: kasprov was it was it? Ken Jennings who played Deep Blue. Yeah,

407
00:24:59,080 --> 00:25:04,879
Speaker 1: it was our own buddy colleague of omnibus fame, and

408
00:25:04,880 --> 00:25:10,000
Speaker 1: then other things. Yeah he does think yeah, yeah, yeah,

409
00:25:10,119 --> 00:25:13,280
Speaker 1: in addition to you know, his big tent item, which

410
00:25:13,320 --> 00:25:17,600
Speaker 1: is hanging hanging out with us. How stuff works. Right, So,

411
00:25:17,680 --> 00:25:20,879
Speaker 1: now we are on the cusp of a world that

412
00:25:21,000 --> 00:25:27,200
Speaker 1: is both brave to quote Watson, Watson, Watson, I said

413
00:25:27,200 --> 00:25:29,360
Speaker 1: Deep Blue, and that's the chess with the chess one.

414
00:25:29,400 --> 00:25:32,360
Speaker 1: Watson was the quiz the quiz bot also by IBM

415
00:25:32,400 --> 00:25:37,880
Speaker 1: though right, okay, yes, so Ken Jennings versus Watson, caspar

416
00:25:37,960 --> 00:25:41,480
Speaker 1: Off versus Deep Blue. And now we are on the

417
00:25:41,560 --> 00:25:46,560
Speaker 1: cusp of a new world that is both brave to

418
00:25:46,640 --> 00:25:51,000
Speaker 1: quote Altis Huxley, and strange. The average citizen in a

419
00:25:51,080 --> 00:25:55,959
Speaker 1: developed country interacts with some form of artificial intelligence on

420
00:25:56,000 --> 00:26:00,320
Speaker 1: an increasingly frequent basis, even if it's indirect. And think

421
00:26:00,320 --> 00:26:02,480
Speaker 1: about it, when is the last time you called a

422
00:26:02,600 --> 00:26:06,200
Speaker 1: large company and didn't first go through an automated line

423
00:26:06,280 --> 00:26:10,399
Speaker 1: A rough impersonation of a conversation with a computer. I

424
00:26:10,520 --> 00:26:14,240
Speaker 1: just always start mashing zero right away just to pass

425
00:26:14,280 --> 00:26:17,840
Speaker 1: them because they it's so annoying, because it's like if

426
00:26:17,920 --> 00:26:21,760
Speaker 1: I thought that, uh, that a computer simulation could help

427
00:26:21,800 --> 00:26:23,720
Speaker 1: me solve the problem. I probably would have just done

428
00:26:23,720 --> 00:26:26,320
Speaker 1: this online. The fact that I'm calling the company in

429
00:26:26,359 --> 00:26:28,320
Speaker 1: the first place not to derail it with my might

430
00:26:28,359 --> 00:26:31,040
Speaker 1: get off my lawn moment. Please do seriously. It's like

431
00:26:31,480 --> 00:26:33,600
Speaker 1: they're never that helpful. You always have to repeat yourself

432
00:26:33,600 --> 00:26:36,200
Speaker 1: over and over again, and you usually do not get

433
00:26:36,200 --> 00:26:40,600
Speaker 1: the result you're looking for. Um, which could change with

434
00:26:41,000 --> 00:26:43,760
Speaker 1: some of the new technology we were talking about today,

435
00:26:43,800 --> 00:26:46,280
Speaker 1: which I think we can get into right now. Possibly

436
00:26:46,680 --> 00:26:49,359
Speaker 1: maybe maybe it will change because one of the big

437
00:26:49,400 --> 00:26:53,400
Speaker 1: task for artificial intelligence right now, what what we use

438
00:26:53,480 --> 00:26:58,040
Speaker 1: these applications for now is big data. We have built

439
00:26:58,080 --> 00:27:01,760
Speaker 1: so many ways to scoop up in formation that are poor.

440
00:27:01,800 --> 00:27:05,399
Speaker 1: Old primate brains, which are built for foraging and living

441
00:27:05,400 --> 00:27:09,080
Speaker 1: in small bands and forests, cannot analyze and process all

442
00:27:09,119 --> 00:27:12,679
Speaker 1: this stuff ourselves, so we built machines and codes. In

443
00:27:12,680 --> 00:27:15,399
Speaker 1: this case, codes could just be standing for lines of

444
00:27:15,480 --> 00:27:18,679
Speaker 1: thought to decipher all this data for us. And the

445
00:27:18,720 --> 00:27:22,360
Speaker 1: application of artificial intelligence in this regard has already done

446
00:27:22,400 --> 00:27:27,560
Speaker 1: amazing things in industries like technology, banking, marketing, entertainments. It's

447
00:27:27,600 --> 00:27:31,280
Speaker 1: automation of like sweeping through giant sets of data so

448
00:27:31,280 --> 00:27:33,919
Speaker 1: that humans don't have to do it, and using particular

449
00:27:34,520 --> 00:27:37,040
Speaker 1: um what's the word algorithms, I guess is one way

450
00:27:37,040 --> 00:27:39,800
Speaker 1: of putting it or just rules saying look for this

451
00:27:39,920 --> 00:27:42,760
Speaker 1: thing in this set of data, parts out this file,

452
00:27:42,840 --> 00:27:45,280
Speaker 1: if then whatever. So that's kind of been the primary

453
00:27:45,359 --> 00:27:47,320
Speaker 1: use of that up to this point in all of

454
00:27:47,320 --> 00:27:51,760
Speaker 1: those things you said, So, um, what's next? Right right?

455
00:27:51,800 --> 00:27:56,280
Speaker 1: Because right now it doesn't matter if whether the algorithms

456
00:27:56,400 --> 00:27:59,840
Speaker 1: improve much, it doesn't matter whether there is a C chain.

457
00:28:00,119 --> 00:28:04,000
Speaker 1: The basic concept is there. And this massive computing approach,

458
00:28:04,080 --> 00:28:06,760
Speaker 1: this vacuum cleaner approach, for instance, that the n s

459
00:28:06,800 --> 00:28:12,240
Speaker 1: A uses, just allows artificial intelligence to learn through brute force.

460
00:28:12,600 --> 00:28:16,280
Speaker 1: In the future, we're going to see more variation in AI.

461
00:28:16,359 --> 00:28:20,360
Speaker 1: We'll see autonomous vehicles, will see predictive ordering services, which

462
00:28:20,400 --> 00:28:23,600
Speaker 1: I know most of us will hate, and soon it

463
00:28:23,640 --> 00:28:27,119
Speaker 1: will seem strange not to have some sort of artificial

464
00:28:27,160 --> 00:28:31,879
Speaker 1: intelligence existing in some aspect of your everyday life. But

465
00:28:32,160 --> 00:28:35,600
Speaker 1: today's question, and a spooky one, is how close to

466
00:28:35,720 --> 00:28:40,320
Speaker 1: human can these programs actually get? And I think we'll

467
00:28:40,400 --> 00:28:43,520
Speaker 1: get to that UM after just one more quick little

468
00:28:43,560 --> 00:28:58,280
Speaker 1: sponsor break. Here's where it gets crazy. Hi, Um, I'd

469
00:28:58,280 --> 00:29:05,800
Speaker 1: like to reserve a table for Wednesday seven for seven people. Um,

470
00:29:05,840 --> 00:29:12,280
Speaker 1: it's for four people? Well people when um next Wednesday

471
00:29:12,440 --> 00:29:17,560
Speaker 1: at six pm? Oh, actually we leave her for like April,

472
00:29:17,680 --> 00:29:20,680
Speaker 1: like a bye people for you? For people, you can come.

473
00:29:22,360 --> 00:29:26,560
Speaker 1: How long is the way usually to be seated? When

474
00:29:26,640 --> 00:29:33,320
Speaker 1: tomorrow or week or four next Wednesday the seventh? Oh no,

475
00:29:33,480 --> 00:29:37,240
Speaker 1: it's not too easy. You can come from all people. Okay,

476
00:29:38,400 --> 00:29:43,400
Speaker 1: oh I got you, thanks, bye bye. But then that

477
00:29:43,360 --> 00:29:45,560
Speaker 1: that that didn't sound like anything at all. That was

478
00:29:45,600 --> 00:29:48,480
Speaker 1: just a conversation between a young man and the proprietor

479
00:29:48,520 --> 00:29:51,600
Speaker 1: of a of a Chinese restaurant, right and who was?

480
00:29:51,720 --> 00:29:57,560
Speaker 1: Who is? Whoever has a great time making restaurant reservation?

481
00:29:57,680 --> 00:29:59,960
Speaker 1: Now you know because when we have open table for

482
00:30:00,000 --> 00:30:03,120
Speaker 1: of that? Am I? Right? Bend? I like many people

483
00:30:03,160 --> 00:30:06,600
Speaker 1: in our generation hate talking on the phone. It's true. Um,

484
00:30:06,600 --> 00:30:08,040
Speaker 1: there's even a term for that. I think was a

485
00:30:08,120 --> 00:30:10,800
Speaker 1: telephone a phobia? Yes, yes, that's a good one. But

486
00:30:10,880 --> 00:30:12,840
Speaker 1: now I felt, yeah, what I found that one I

487
00:30:12,960 --> 00:30:15,040
Speaker 1: was like, how can you make a can you just

488
00:30:15,080 --> 00:30:17,320
Speaker 1: make anything of phobia? And I you can. I think

489
00:30:17,320 --> 00:30:20,360
Speaker 1: it's very American English. Now I have microphone a phobia,

490
00:30:20,400 --> 00:30:22,440
Speaker 1: which is weird considering that I sit in front of

491
00:30:22,440 --> 00:30:25,320
Speaker 1: one all the time. Right now, I'm in utter terror.

492
00:30:25,360 --> 00:30:27,960
Speaker 1: But no, I was being coy. That clip was not

493
00:30:28,160 --> 00:30:31,640
Speaker 1: between two humans. I think you can probably guess which

494
00:30:31,640 --> 00:30:35,239
Speaker 1: side of the conversation was the non human, because if

495
00:30:35,240 --> 00:30:37,800
Speaker 1: you listen back to it, I think you'll notice a

496
00:30:37,800 --> 00:30:41,720
Speaker 1: couple of tells um and that there was a few

497
00:30:42,120 --> 00:30:44,320
Speaker 1: There were a few kind of wrenches thrown into that

498
00:30:44,320 --> 00:30:48,120
Speaker 1: that exchange where the person was trying to call to

499
00:30:48,120 --> 00:30:51,640
Speaker 1: make a reservation, and there was some confusion from the

500
00:30:51,720 --> 00:30:54,080
Speaker 1: person on the other end about how many, about what

501
00:30:54,160 --> 00:30:56,880
Speaker 1: day they wanted the reservation to be, how many people

502
00:30:56,920 --> 00:31:00,520
Speaker 1: were in the party, and it ultimately ended up with, well,

503
00:31:00,560 --> 00:31:02,920
Speaker 1: you don't really need a reservation. So the person on

504
00:31:02,960 --> 00:31:04,600
Speaker 1: the other on the phone was not equipped. It was

505
00:31:04,600 --> 00:31:07,000
Speaker 1: not something they would typically do. Reservations seemed like not

506
00:31:07,040 --> 00:31:09,880
Speaker 1: really a thing in this restaurant, and so the caller

507
00:31:11,520 --> 00:31:16,440
Speaker 1: kind of repeated himself specific things like the specific time

508
00:31:16,800 --> 00:31:20,000
Speaker 1: and day, and then there was a momentary pause when

509
00:31:20,000 --> 00:31:22,840
Speaker 1: they said we we only do reservations for groups of

510
00:31:22,960 --> 00:31:27,960
Speaker 1: five or more, gotcha, which still sounded very human did

511
00:31:28,160 --> 00:31:30,520
Speaker 1: it was not though it was a conversation between an

512
00:31:30,560 --> 00:31:34,760
Speaker 1: unsuspecting restaurant employee, as you mentioned, and a computer program.

513
00:31:34,800 --> 00:31:40,880
Speaker 1: This is an example of Google's Duplex system, a personal

514
00:31:40,880 --> 00:31:45,720
Speaker 1: assistant designed to make users lives easier by handling standard

515
00:31:45,720 --> 00:31:49,880
Speaker 1: phone calls, so like doctors, appointments, reservations, and so on,

516
00:31:50,720 --> 00:31:54,120
Speaker 1: breaking up with a loved one without troubling you or

517
00:31:54,240 --> 00:31:56,920
Speaker 1: possibly even letting the person on the other end of

518
00:31:56,960 --> 00:32:00,240
Speaker 1: the phone know that it's not actually you making the call. Yeah,

519
00:32:00,280 --> 00:32:02,440
Speaker 1: And we get into all kinds of ethical quandaries with

520
00:32:02,480 --> 00:32:04,800
Speaker 1: this that we'll get into first first and foremost, though,

521
00:32:04,840 --> 00:32:08,960
Speaker 1: I think it's interesting that the idea, it's it's inherently tricksy,

522
00:32:09,160 --> 00:32:12,400
Speaker 1: the whole the whole affair, right. The idea is to

523
00:32:13,600 --> 00:32:17,360
Speaker 1: not inconvenience either party. But you are you are genuinely

524
00:32:17,440 --> 00:32:20,320
Speaker 1: kind of tricking someone into believing they're talking to a person. Yes,

525
00:32:20,480 --> 00:32:22,920
Speaker 1: and you're giving up a lot of stuff. So at

526
00:32:22,960 --> 00:32:28,120
Speaker 1: first this seems amazing because, like we established earlier, telephone

527
00:32:28,160 --> 00:32:30,840
Speaker 1: and phobia, in addition to being really fun to say,

528
00:32:31,080 --> 00:32:34,840
Speaker 1: is a genuine thing. And I would argue increasingly people

529
00:32:34,880 --> 00:32:37,800
Speaker 1: are resorting to text. I mean, how often do you

530
00:32:37,840 --> 00:32:39,080
Speaker 1: talk on the phone if you're not talking to like

531
00:32:39,080 --> 00:32:43,000
Speaker 1: your mom or dad, like on a preappointed time. I mean,

532
00:32:43,080 --> 00:32:46,080
Speaker 1: if I'm doing interviews for a podcast, but that's a

533
00:32:46,200 --> 00:32:49,680
Speaker 1: very that's a very specific task, not for fun, never

534
00:32:49,760 --> 00:32:52,200
Speaker 1: for fun, and unless I just feel like I could

535
00:32:52,200 --> 00:32:54,240
Speaker 1: suck something out a lot quicker on the phone, real

536
00:32:54,320 --> 00:32:57,640
Speaker 1: quick than I could usually for business though, for work stuff,

537
00:32:57,640 --> 00:33:01,160
Speaker 1: because sometimes or conference calls, the dreaded ofference call. That's

538
00:33:01,280 --> 00:33:03,920
Speaker 1: also works thing that still happens. But yeah, for fun,

539
00:33:04,400 --> 00:33:08,440
Speaker 1: not very often, um, you know, but for other people

540
00:33:08,520 --> 00:33:11,720
Speaker 1: in the mix. There's totally a dark side to this

541
00:33:11,840 --> 00:33:16,400
Speaker 1: idea of the super convenient, amazing idea of my phone

542
00:33:16,440 --> 00:33:19,000
Speaker 1: being able to make calls for me and set up

543
00:33:19,000 --> 00:33:23,880
Speaker 1: my you know, my waxing appointment or whatever. Because the

544
00:33:23,920 --> 00:33:27,479
Speaker 1: implications are incredibly far reaching, aren't they, ben Yeah, they

545
00:33:27,480 --> 00:33:31,840
Speaker 1: are old. It's not just a it's not just an

546
00:33:31,880 --> 00:33:37,520
Speaker 1: automated phone line saying press one if you like to

547
00:33:37,560 --> 00:33:40,480
Speaker 1: make a payment, press two if you'd like to hear

548
00:33:40,800 --> 00:33:44,720
Speaker 1: mailing address. It's not just impersoning a generic human now,

549
00:33:44,760 --> 00:33:49,920
Speaker 1: it's a program capable of impersonating specific humans, namely you

550
00:33:50,280 --> 00:33:53,920
Speaker 1: capital y oh, you specifically you listening to this, and

551
00:33:53,920 --> 00:33:55,960
Speaker 1: it's got a lot of it's got a lot of

552
00:33:55,960 --> 00:34:00,400
Speaker 1: opponents already. Now we're not saying that it it mimics

553
00:34:00,440 --> 00:34:03,880
Speaker 1: your voice. That's not what this is implying, not yet,

554
00:34:04,080 --> 00:34:06,320
Speaker 1: not yet. But the idea is that it is. It

555
00:34:06,360 --> 00:34:10,080
Speaker 1: has access to all your information. It has access to

556
00:34:10,160 --> 00:34:13,640
Speaker 1: your date books and your phone numbers, and and knows

557
00:34:13,680 --> 00:34:16,799
Speaker 1: everything about you because it's tied into this little thing

558
00:34:16,840 --> 00:34:19,240
Speaker 1: you carry around with you that is basically your life

559
00:34:19,400 --> 00:34:23,040
Speaker 1: in a wallet form you know, more specifically with Google.

560
00:34:23,520 --> 00:34:27,719
Speaker 1: Uh this, this initial fear increases, or the potential for

561
00:34:27,800 --> 00:34:31,480
Speaker 1: misuse increases. This journalist who are an excellent article and

562
00:34:31,600 --> 00:34:35,680
Speaker 1: her name's Alex Krans to her, the dangers are threefold. First,

563
00:34:36,040 --> 00:34:39,200
Speaker 1: similar to what what you're proposing, all this is a

564
00:34:39,239 --> 00:34:44,399
Speaker 1: program from Google from Alphabet, and Google already knows a

565
00:34:44,400 --> 00:34:47,560
Speaker 1: ton of personal information about you, whether or not you

566
00:34:47,760 --> 00:34:50,759
Speaker 1: use an Android phone. This means that Duplex doesn't just

567
00:34:50,920 --> 00:34:53,640
Speaker 1: know what time you want to meet your tender date

568
00:34:53,680 --> 00:34:57,319
Speaker 1: at that tapas place. It also potentially knows every other

569
00:34:57,400 --> 00:35:01,440
Speaker 1: single thing Google knows about your life. And Google is

570
00:35:01,480 --> 00:35:04,000
Speaker 1: so widespread that you do not have to have a

571
00:35:04,000 --> 00:35:07,440
Speaker 1: Google account for this to touch you. Second, a criminal,

572
00:35:07,600 --> 00:35:13,680
Speaker 1: a government stalker, prankster, some internet roll could potentially cheat

573
00:35:13,800 --> 00:35:17,719
Speaker 1: Duplex out of giving up your information in a phone call.

574
00:35:18,080 --> 00:35:22,279
Speaker 1: Imagine Duplex somehow being fooled into thinking it's calling a

575
00:35:22,400 --> 00:35:26,600
Speaker 1: restaurant to make a delivery order, and then boom, somebody

576
00:35:26,640 --> 00:35:29,960
Speaker 1: has your credit card information, expiration date, and security code.

577
00:35:30,440 --> 00:35:33,399
Speaker 1: They're actually a really great a whole bunch of great

578
00:35:33,440 --> 00:35:36,200
Speaker 1: comments in these in the comment section on this Gizmodo

579
00:35:36,360 --> 00:35:39,120
Speaker 1: article by Alex Krans, and one of them was the

580
00:35:39,160 --> 00:35:41,920
Speaker 1: idea of, like how this technology could be used by

581
00:35:42,160 --> 00:35:45,400
Speaker 1: this term I love bad actors, you know, So for example,

582
00:35:45,480 --> 00:35:48,840
Speaker 1: like what if you had a pizza place that wanted

583
00:35:48,880 --> 00:35:53,680
Speaker 1: to use this kind of technology to flood arrival pizza

584
00:35:53,719 --> 00:35:57,640
Speaker 1: company with fake calls and keep their phones tied up

585
00:35:57,680 --> 00:35:59,719
Speaker 1: and and I don't know that was the particular one

586
00:35:59,719 --> 00:36:02,640
Speaker 1: and sort far feeshied one, but the idea, the implications

587
00:36:02,719 --> 00:36:08,000
Speaker 1: being that you could use this for scamming easily, big time,

588
00:36:08,280 --> 00:36:11,399
Speaker 1: especially with older folks that maybe aren't going to key

589
00:36:11,440 --> 00:36:13,480
Speaker 1: into the fact that this is not a real person,

590
00:36:13,880 --> 00:36:18,759
Speaker 1: and you could possibly automate and scale scamming in a

591
00:36:18,800 --> 00:36:21,239
Speaker 1: way that would far surpass the way it's done now,

592
00:36:21,320 --> 00:36:24,360
Speaker 1: because even the least tech savvy people can tell that

593
00:36:24,440 --> 00:36:28,200
Speaker 1: you're hearing a automated voice, you know, spamming you to

594
00:36:28,440 --> 00:36:30,840
Speaker 1: try to you know, sign up for a cruise or

595
00:36:30,840 --> 00:36:35,239
Speaker 1: something like that. Absolutely, And and to follow up on

596
00:36:35,520 --> 00:36:37,440
Speaker 1: Crans is because you said it was threefold, So to

597
00:36:37,520 --> 00:36:40,680
Speaker 1: follow up on the third thing, which I found the

598
00:36:41,040 --> 00:36:45,839
Speaker 1: perhaps the most disturbing here, No forget, perhaps it is

599
00:36:45,880 --> 00:36:50,239
Speaker 1: the most disturbing thing. A person could hijack your duplex

600
00:36:50,360 --> 00:36:55,640
Speaker 1: account and essentially function as you until such a point

601
00:36:55,719 --> 00:36:58,840
Speaker 1: as they're caught. So imagine learning your credit and savings

602
00:36:58,840 --> 00:37:01,600
Speaker 1: have been wiped out because you're duplex called in a

603
00:37:01,640 --> 00:37:06,280
Speaker 1: series of untraceable transfers to some Caribbean islands, some offshore account.

604
00:37:06,680 --> 00:37:10,279
Speaker 1: There's nothing that anyone could do because legally that thing

605
00:37:10,320 --> 00:37:13,600
Speaker 1: would be functioning as you. Crans and other critics of

606
00:37:13,640 --> 00:37:17,480
Speaker 1: this technology, you're also concerned that at the initial presentation,

607
00:37:18,200 --> 00:37:21,239
Speaker 1: Google build this as another Needo feature and they didn't

608
00:37:21,239 --> 00:37:23,919
Speaker 1: say anything about privacy or concerns. It was very much,

609
00:37:23,960 --> 00:37:27,719
Speaker 1: look what we can do, not hey should we just

610
00:37:27,800 --> 00:37:30,080
Speaker 1: like like it's like that line from Jurassic Park. Your

611
00:37:30,480 --> 00:37:33,040
Speaker 1: your scientists were spending so much time thinking whether they could.

612
00:37:33,280 --> 00:37:37,680
Speaker 1: They never thought if they should, and crime finds a way.

613
00:37:37,760 --> 00:37:40,840
Speaker 1: It does. It does. But but you know, and you

614
00:37:40,880 --> 00:37:42,839
Speaker 1: may think this is an alarmist, but it's just like,

615
00:37:42,920 --> 00:37:44,680
Speaker 1: we're just a hop, skip and a jump away from

616
00:37:44,680 --> 00:37:47,920
Speaker 1: this stuff. And obviously this technology is not ready for marketing. No,

617
00:37:48,080 --> 00:37:49,920
Speaker 1: no one's saying that this is just kind of like

618
00:37:49,960 --> 00:37:53,280
Speaker 1: a Needo party trick they did at their io Developers conference.

619
00:37:53,480 --> 00:37:55,720
Speaker 1: But yeah, you're right, because if if, if it's literally

620
00:37:55,840 --> 00:38:00,840
Speaker 1: the voice inside your phone, then it's going to have

621
00:38:00,880 --> 00:38:03,279
Speaker 1: access to all the data that's in your phone. And

622
00:38:03,360 --> 00:38:06,000
Speaker 1: depending on how careful you are with this stuff, it

623
00:38:06,040 --> 00:38:09,160
Speaker 1: could have every bit of identifying information that a human

624
00:38:09,280 --> 00:38:13,480
Speaker 1: banker would ask you to give to confirm this crazy

625
00:38:13,520 --> 00:38:14,960
Speaker 1: wire transfer that you want to do. What do they

626
00:38:15,000 --> 00:38:17,200
Speaker 1: ask you for? Ben they ask you for like the

627
00:38:17,280 --> 00:38:20,840
Speaker 1: last four digits of your social Security number, your mailing address,

628
00:38:20,840 --> 00:38:24,000
Speaker 1: your name, maybe your mother's maiden name, or some secret word.

629
00:38:25,040 --> 00:38:27,920
Speaker 1: Questions gonna be on your phone in some way, shape

630
00:38:28,000 --> 00:38:31,799
Speaker 1: or form likely, and many people email themselves those answers

631
00:38:31,840 --> 00:38:35,160
Speaker 1: so that when they're online they can check back, including passwords.

632
00:38:35,440 --> 00:38:38,600
Speaker 1: So it's a very important point to make that it

633
00:38:38,680 --> 00:38:41,399
Speaker 1: has this duplex stuff has not been assigned any kind

634
00:38:41,400 --> 00:38:44,720
Speaker 1: of rollout date. People are still speculating just how close

635
00:38:44,840 --> 00:38:48,680
Speaker 1: or how far Google is from making this technology viable

636
00:38:48,760 --> 00:38:52,000
Speaker 1: outside of testing situations and later available to the public.

637
00:38:52,200 --> 00:38:54,600
Speaker 1: But the problem is this is not the only game

638
00:38:54,600 --> 00:38:59,240
Speaker 1: in town. Sure, audio impersonation is spooky, but what about video.

639
00:38:59,360 --> 00:39:02,319
Speaker 1: Do you remember back during the hunt for Osama bin

640
00:39:02,400 --> 00:39:06,040
Speaker 1: Laden where various people were arguing that the bin Laden

641
00:39:06,040 --> 00:39:09,399
Speaker 1: and propaganda videos wasn't the real guy, but someone else

642
00:39:09,440 --> 00:39:13,359
Speaker 1: impersonating him or something like that, or like a Snapchat filter. Right,

643
00:39:13,600 --> 00:39:15,920
Speaker 1: he's kidding, but I'm not because all of the crazy

644
00:39:16,000 --> 00:39:19,080
Speaker 1: tech blows my mind. How much research and development they

645
00:39:19,080 --> 00:39:21,799
Speaker 1: put into the Snapchat filters, And the more you see them,

646
00:39:21,840 --> 00:39:23,719
Speaker 1: the more realistic they get, and the more I could

647
00:39:23,760 --> 00:39:27,000
Speaker 1: see them turning into some pretty nefarious ways of mapping

648
00:39:27,040 --> 00:39:29,719
Speaker 1: people's faces and you know, making you look like someone

649
00:39:29,760 --> 00:39:32,360
Speaker 1: else entirely and making it not just a silly mustache

650
00:39:32,760 --> 00:39:37,640
Speaker 1: but a totally believable doppel gang. Yeah, and we've seen Look,

651
00:39:38,280 --> 00:39:41,640
Speaker 1: visual manipulation is a tale as old as journalism. We've

652
00:39:41,640 --> 00:39:46,600
Speaker 1: seen doctored photos of plenty purporting to be evidence of UFOs, giants,

653
00:39:46,600 --> 00:39:50,680
Speaker 1: skeleton's ghosts, and we has a species. Have known for

654
00:39:50,680 --> 00:39:53,680
Speaker 1: a long time that photos can be faked. Someone with

655
00:39:53,719 --> 00:39:59,400
Speaker 1: a handy knack for photoshop, Paul can really work wonders,

656
00:39:59,480 --> 00:40:02,200
Speaker 1: and now days we've gone beyond the touch ups of

657
00:40:02,320 --> 00:40:04,640
Speaker 1: still images. Have you seen that doctor and stuff? They

658
00:40:04,640 --> 00:40:08,560
Speaker 1: don't want you to know? Um logo, there's some something

659
00:40:08,600 --> 00:40:12,800
Speaker 1: fishy gone on with that hand suspicious indeed, but again

660
00:40:12,840 --> 00:40:17,360
Speaker 1: no spoilers, right, So, uh, if you're wondering fans of

661
00:40:17,400 --> 00:40:20,640
Speaker 1: West World, yes, that Mays is for you. As The

662
00:40:20,719 --> 00:40:24,200
Speaker 1: Verge reported in July of last year, seen we're recording

663
00:40:24,200 --> 00:40:28,680
Speaker 1: this July, researchers at the University of Washington invented a

664
00:40:28,719 --> 00:40:33,320
Speaker 1: tool that takes audio files, converts them to realistic mouth movements,

665
00:40:33,480 --> 00:40:37,000
Speaker 1: and then graphs those onto existing video. The end result

666
00:40:37,080 --> 00:40:39,839
Speaker 1: of this is someone saying something or appearing to say

667
00:40:39,880 --> 00:40:42,560
Speaker 1: something that they never actually said. And the scary part

668
00:40:42,640 --> 00:40:45,759
Speaker 1: is it's really convincing. It's not like meme level give

669
00:40:46,000 --> 00:40:49,359
Speaker 1: Funny a lot of their earlier examples used footage of

670
00:40:49,719 --> 00:40:52,760
Speaker 1: former President Barack Obama. Did you see this stuff? Well, dude,

671
00:40:52,760 --> 00:40:54,600
Speaker 1: there I mean, and I don't not keep harping on this,

672
00:40:54,719 --> 00:40:59,400
Speaker 1: But there was a Barack Obama Snapchat filter and it

673
00:40:59,400 --> 00:41:02,759
Speaker 1: it does have that Uncanny Valley look. But first of all,

674
00:41:02,800 --> 00:41:06,400
Speaker 1: problematic for a lot of reasons to put on the

675
00:41:06,719 --> 00:41:09,799
Speaker 1: visage of someone that's a different race than you and

676
00:41:09,960 --> 00:41:13,400
Speaker 1: play a character, and you know, just Snapchat users were

677
00:41:13,480 --> 00:41:15,879
Speaker 1: using it. It was literally a Snapchat field. There's also

678
00:41:15,880 --> 00:41:18,560
Speaker 1: another one that was Bob Marley, you know, with like

679
00:41:18,640 --> 00:41:22,560
Speaker 1: but but it's pretty damn realistic. Other than there's things

680
00:41:22,600 --> 00:41:25,160
Speaker 1: they probably did on purpose to make it less realistic.

681
00:41:25,200 --> 00:41:28,120
Speaker 1: But that kind of mapping quality that you're talking about,

682
00:41:28,440 --> 00:41:33,160
Speaker 1: the implications there are nutty. Yeah, And luckily for everybody

683
00:41:33,239 --> 00:41:37,680
Speaker 1: just got spooped out, we have some good good news here. First,

684
00:41:38,239 --> 00:41:41,840
Speaker 1: they didn't just choose Barack Obama because they had some

685
00:41:41,920 --> 00:41:44,279
Speaker 1: sort of ideological thing or they were like we like

686
00:41:44,440 --> 00:41:46,919
Speaker 1: this guy, we hate this guy, or anything. They did

687
00:41:46,960 --> 00:41:49,840
Speaker 1: it because a high profile individual like a celebrity or

688
00:41:49,920 --> 00:41:52,680
Speaker 1: president will have a ton more high quality video and

689
00:41:52,760 --> 00:41:56,600
Speaker 1: audio footage to pull from. And also, this is this

690
00:41:56,920 --> 00:42:01,080
Speaker 1: thing takes uh, it's a huge wish in process. They

691
00:42:01,080 --> 00:42:04,080
Speaker 1: had to have at least seventeen hours of footage just

692
00:42:04,160 --> 00:42:07,120
Speaker 1: to get started and they say, so there's another neural

693
00:42:07,200 --> 00:42:10,480
Speaker 1: net kind of learning exactly, learning stuffing, So this is

694
00:42:10,520 --> 00:42:14,880
Speaker 1: a problem. They say that their goals are wholly good hearted.

695
00:42:15,120 --> 00:42:17,920
Speaker 1: I've got a quote here. The team behind the works

696
00:42:17,960 --> 00:42:20,240
Speaker 1: say they hope it could be used to improve video

697
00:42:20,320 --> 00:42:24,520
Speaker 1: chat tools like Skype. Users could collect footage of themselves speaking,

698
00:42:25,080 --> 00:42:27,920
Speaker 1: use use it to train the software, and then when

699
00:42:27,920 --> 00:42:30,400
Speaker 1: they need to talk to someone video on their side

700
00:42:30,400 --> 00:42:34,400
Speaker 1: would be generated automatically, just using their voice. This would

701
00:42:34,400 --> 00:42:37,560
Speaker 1: help in situations where someone's Internet is shaky, or if

702
00:42:37,560 --> 00:42:40,760
Speaker 1: they're trying to save mobile data. But that's what Google

703
00:42:40,800 --> 00:42:43,480
Speaker 1: is saying about Duplex. It's just a nifty, little handy

704
00:42:43,560 --> 00:42:46,960
Speaker 1: dandy tool to help you book your hair appointment so

705
00:42:47,040 --> 00:42:48,760
Speaker 1: you don't have to talk to people on the phone.

706
00:42:49,040 --> 00:42:52,520
Speaker 1: They never talk about the yeah, and this, yeah exactly

707
00:42:52,560 --> 00:42:56,120
Speaker 1: and this look. We can't we can't ascribe motive. We

708
00:42:56,160 --> 00:42:59,400
Speaker 1: can't say these people are lying to you, but we

709
00:42:59,520 --> 00:43:03,399
Speaker 1: can a that this sort of technology is a Pandora's jar,

710
00:43:03,640 --> 00:43:07,760
Speaker 1: and once that lid is unscrewed, there's no there's absolutely

711
00:43:07,800 --> 00:43:11,480
Speaker 1: no realistic way to prevent both the spread of these

712
00:43:11,520 --> 00:43:15,920
Speaker 1: faked segments and the spread of things being accused unjustly

713
00:43:16,000 --> 00:43:20,279
Speaker 1: of fake segments. This this poses some inherent dangers for journalism.

714
00:43:20,480 --> 00:43:23,480
Speaker 1: We already see how easily a completely fake story can

715
00:43:23,520 --> 00:43:27,680
Speaker 1: proliferate on Facebook, a bot generated story at times, like

716
00:43:27,719 --> 00:43:32,000
Speaker 1: literally an AI generated Twitter account, for example, that can

717
00:43:32,120 --> 00:43:36,640
Speaker 1: mimic the style of someone of a particular ideology or whatever.

718
00:43:37,200 --> 00:43:40,120
Speaker 1: Um here's another aspect of it. There's an article from

719
00:43:40,160 --> 00:43:44,239
Speaker 1: the information dot Com. Google's controversial voice assistant could talk

720
00:43:44,280 --> 00:43:48,680
Speaker 1: its way into call centers because, like we're saying earlier,

721
00:43:49,440 --> 00:43:51,800
Speaker 1: if you if you get an automated voice on the phone,

722
00:43:52,000 --> 00:43:54,120
Speaker 1: you know it's not going to be actually very helpful

723
00:43:54,160 --> 00:43:56,960
Speaker 1: at all. You're probably just gonna blast past it. And

724
00:43:57,000 --> 00:43:58,960
Speaker 1: they know that. And there are a lot of people

725
00:43:59,040 --> 00:44:01,839
Speaker 1: that have their jaws being that person that it gets

726
00:44:01,880 --> 00:44:06,359
Speaker 1: passed off too. So if there were a more successful

727
00:44:06,960 --> 00:44:10,239
Speaker 1: voice recognition and communication tool like this, it could put

728
00:44:10,280 --> 00:44:12,600
Speaker 1: a lot of people out of jobs. That's a very

729
00:44:12,600 --> 00:44:14,920
Speaker 1: good point. It could get us closer to not the

730
00:44:14,960 --> 00:44:19,200
Speaker 1: post work economy, but the post worker economy. And up

731
00:44:19,200 --> 00:44:23,680
Speaker 1: to now we've talked about these two forms of impersonation

732
00:44:23,800 --> 00:44:27,520
Speaker 1: is discreeting different things, right, audio on one side, video

733
00:44:27,560 --> 00:44:30,359
Speaker 1: on another. But what if they become combined. What if

734
00:44:30,400 --> 00:44:34,279
Speaker 1: a digital impersonation of a human being, even you Know

735
00:44:34,520 --> 00:44:37,440
Speaker 1: or even you Paul, or even me, could exist online

736
00:44:37,520 --> 00:44:40,000
Speaker 1: with no one but you and the people you meet

737
00:44:40,080 --> 00:44:43,319
Speaker 1: in person knowing the difference. I mean, let's think about it.

738
00:44:43,320 --> 00:44:46,040
Speaker 1: It would sound like us, it would look like us,

739
00:44:46,320 --> 00:44:50,160
Speaker 1: and if it pulled from our online data footprint, it

740
00:44:50,200 --> 00:44:52,640
Speaker 1: would also know a lot of stuff about us, including

741
00:44:52,880 --> 00:44:56,040
Speaker 1: the relationships we already have with other people and how

742
00:44:56,040 --> 00:45:00,600
Speaker 1: we interact with them. So it's possible, for instance, that uh,

743
00:45:00,600 --> 00:45:06,440
Speaker 1: fake version of Matt Frederick rights to us from Massachusetts

744
00:45:07,000 --> 00:45:11,319
Speaker 1: foreshadowing there and responds to the three of us the

745
00:45:11,400 --> 00:45:13,880
Speaker 1: way that the real Matt would, and we wouldn't know

746
00:45:13,920 --> 00:45:17,160
Speaker 1: the difference. We'd be sending weird memes and thumbs up

747
00:45:17,160 --> 00:45:19,640
Speaker 1: and doing our inside jokes, and it would already know

748
00:45:19,760 --> 00:45:21,560
Speaker 1: all of those. And let's be honest, I mean, a

749
00:45:21,560 --> 00:45:24,520
Speaker 1: lot of times text conversation and email conversations are already

750
00:45:24,560 --> 00:45:26,640
Speaker 1: kind of in shorthand or in some kind of a

751
00:45:26,640 --> 00:45:28,560
Speaker 1: little bit more terse, not mean, but just kind of

752
00:45:28,600 --> 00:45:30,200
Speaker 1: like we're trying to get it done. That's why we

753
00:45:30,640 --> 00:45:33,640
Speaker 1: depend on that method of communication over talking, because it's

754
00:45:33,680 --> 00:45:36,080
Speaker 1: a lot more boom boom boom, let's get it done

755
00:45:36,080 --> 00:45:37,960
Speaker 1: and move on. I would think it would be easy

756
00:45:38,320 --> 00:45:41,799
Speaker 1: ish for an AI to mimic those kinds of little

757
00:45:41,880 --> 00:45:44,840
Speaker 1: quick exchanges and not raise a red flag for you

758
00:45:44,960 --> 00:45:46,799
Speaker 1: or I. That's a great point, because even if they

759
00:45:46,840 --> 00:45:48,879
Speaker 1: get something wrong, we would just think, oh, Matt must

760
00:45:48,920 --> 00:45:51,960
Speaker 1: be in a hurry typo or you know, fat finger whatever,

761
00:45:52,680 --> 00:45:56,879
Speaker 1: and or auto correct correct exactly like I've I've I've

762
00:45:56,880 --> 00:45:59,880
Speaker 1: sent texts and like seeing how mangled they were, but

763
00:46:00,080 --> 00:46:02,040
Speaker 1: be like, oh, he'll know what I mean. Yes, Like

764
00:46:02,160 --> 00:46:04,560
Speaker 1: let me let it ride, yeah, yeah, yeah, because again

765
00:46:04,600 --> 00:46:06,880
Speaker 1: it goes the time. Now, it would be alarmist for

766
00:46:06,960 --> 00:46:10,120
Speaker 1: us to say this would plausibly happen to the average

767
00:46:10,200 --> 00:46:13,320
Speaker 1: person in the near future, but for a high value

768
00:46:13,360 --> 00:46:17,600
Speaker 1: target like a politician, a celebrity, controversial business person and

769
00:46:17,640 --> 00:46:20,719
Speaker 1: so on, it's completely within the realm of possibility. And

770
00:46:20,760 --> 00:46:23,920
Speaker 1: speaking of alarmists, let's open wide the doors of science

771
00:46:23,960 --> 00:46:27,759
Speaker 1: fiction and do a bit of speculation here, completely unfounded.

772
00:46:28,040 --> 00:46:31,200
Speaker 1: Imagine a world where the only communication you can trust

773
00:46:31,480 --> 00:46:35,919
Speaker 1: becomes face to face in person. Imagine a world where

774
00:46:36,000 --> 00:46:38,600
Speaker 1: you are accused of a crime you didn't commit, but

775
00:46:38,800 --> 00:46:41,920
Speaker 1: you don't have a rock solid alibi of your activities

776
00:46:41,960 --> 00:46:44,560
Speaker 1: at the time of the alleged crime, and surprise, surprise,

777
00:46:44,800 --> 00:46:47,480
Speaker 1: they have you on camera committing it and then confessing

778
00:46:47,480 --> 00:46:49,640
Speaker 1: to it, and you cannot prove it is not you.

779
00:46:49,880 --> 00:46:53,920
Speaker 1: Would this mean that eventually video evidence becomes inadmissible in court?

780
00:46:54,200 --> 00:46:57,520
Speaker 1: Does this technology pose an existential threat to the fabric

781
00:46:57,600 --> 00:47:02,000
Speaker 1: of digital reality? Is it really us talking to you

782
00:47:02,440 --> 00:47:05,320
Speaker 1: right now? Yeah? Man, I don't even know anymore. I

783
00:47:05,400 --> 00:47:09,799
Speaker 1: don't know. And and to to answer your hypothetical question earlier, yes,

784
00:47:09,880 --> 00:47:12,080
Speaker 1: I think this is absolutely a threat, and we rely

785
00:47:12,280 --> 00:47:14,719
Speaker 1: on things like video evidence. But I could see far

786
00:47:14,840 --> 00:47:18,719
Speaker 1: enough removed from this particular technological time and place that

787
00:47:18,800 --> 00:47:21,399
Speaker 1: could be something that is just a thing of the past. Man,

788
00:47:21,520 --> 00:47:23,880
Speaker 1: you know what I mean? Can can can you picture it?

789
00:47:23,920 --> 00:47:25,680
Speaker 1: Like I'm trying to think of a of an analog

790
00:47:25,760 --> 00:47:29,000
Speaker 1: of something that used to be infallible and now is

791
00:47:29,000 --> 00:47:32,799
Speaker 1: like completely up for grabs, like I don't know, polygraphs, Yeah,

792
00:47:32,840 --> 00:47:35,320
Speaker 1: polygraphs exactly a great example. I don't know. This is

793
00:47:35,320 --> 00:47:37,399
Speaker 1: a little bit more of an ephemeral example. But even

794
00:47:37,400 --> 00:47:39,600
Speaker 1: something like the press or like the news you know

795
00:47:39,960 --> 00:47:41,920
Speaker 1: now you know, used to to be able to depend

796
00:47:42,000 --> 00:47:44,960
Speaker 1: on some level of rigor and truth in any kind

797
00:47:44,960 --> 00:47:47,000
Speaker 1: of news reporting you see. And now, as you say,

798
00:47:47,000 --> 00:47:49,600
Speaker 1: because of the Internet, we can see stuff that's completely

799
00:47:50,320 --> 00:47:55,080
Speaker 1: generated by artificial intelligence that tricks people all the time

800
00:47:55,080 --> 00:47:58,600
Speaker 1: into believing that it's you know, God's truth, and it's insane.

801
00:47:58,680 --> 00:48:02,920
Speaker 1: So I wonder you know, I have questions for you

802
00:48:03,040 --> 00:48:05,000
Speaker 1: to know. I don't want to put you on the spot,

803
00:48:05,040 --> 00:48:08,239
Speaker 1: but this is this is one of the only conversations

804
00:48:08,280 --> 00:48:10,840
Speaker 1: we can know is actually happening between us, right, I

805
00:48:10,880 --> 00:48:12,480
Speaker 1: mean we are sitting here and looking at each other

806
00:48:12,560 --> 00:48:15,600
Speaker 1: in the eyes, so we are definitely both humans because

807
00:48:15,600 --> 00:48:20,520
Speaker 1: as far as I know, we don't have human stand

808
00:48:20,560 --> 00:48:22,920
Speaker 1: ins that are believable enough to to trick either one

809
00:48:22,920 --> 00:48:25,200
Speaker 1: of us. It's like in Terminator, the first ones you

810
00:48:25,239 --> 00:48:29,160
Speaker 1: could tell the difference, right, So, so what do you

811
00:48:29,239 --> 00:48:32,600
Speaker 1: have any thoughts on the likelihood of this kind of

812
00:48:32,640 --> 00:48:37,520
Speaker 1: technology progressing or being used to disrupt the spread of

813
00:48:37,680 --> 00:48:40,480
Speaker 1: reliable information. Do you think it's a definite Do you

814
00:48:40,480 --> 00:48:43,440
Speaker 1: think it might happen. Do you think it's a little

815
00:48:43,440 --> 00:48:46,920
Speaker 1: bit sensationalistic? What do you think? I think it absolutely

816
00:48:46,920 --> 00:48:49,160
Speaker 1: could happen. I think it's another one of those things

817
00:48:49,239 --> 00:48:54,120
Speaker 1: that's kind of um dangled out there like a piece

818
00:48:54,160 --> 00:48:57,080
Speaker 1: of bait on a fishing hook for the American consumer

819
00:48:57,120 --> 00:48:59,719
Speaker 1: to bite onto and be, oh, this will change my

820
00:48:59,800 --> 00:49:01,760
Speaker 1: life for the better. This will make it so easy

821
00:49:01,840 --> 00:49:04,560
Speaker 1: for me to not have to make hair appointments or

822
00:49:04,800 --> 00:49:08,160
Speaker 1: schedule oil changes or what have you. And then you,

823
00:49:08,160 --> 00:49:10,600
Speaker 1: you know, once the buy in happens, then it started.

824
00:49:10,719 --> 00:49:13,400
Speaker 1: It opens that Pandora's jar that you keep talking about.

825
00:49:13,560 --> 00:49:15,359
Speaker 1: So yeah, no, there's no doubt about it. The more

826
00:49:15,400 --> 00:49:18,680
Speaker 1: access we give our devices and the more centralized our

827
00:49:18,719 --> 00:49:21,719
Speaker 1: information is. And I guess by centralized, I just mean

828
00:49:21,800 --> 00:49:26,080
Speaker 1: within a particular service where it has access to all

829
00:49:26,080 --> 00:49:28,600
Speaker 1: of it um. And that's that's what That's what it takes. Right,

830
00:49:29,120 --> 00:49:31,040
Speaker 1: Why does series suck? Okay, let'ten. I just want to

831
00:49:31,040 --> 00:49:32,600
Speaker 1: get into this because I want to rag on Apple

832
00:49:32,640 --> 00:49:35,680
Speaker 1: for a minute. Siri is not a good personal assistant

833
00:49:35,920 --> 00:49:40,040
Speaker 1: because it doesn't learn, it doesn't keep track of what

834
00:49:40,080 --> 00:49:42,960
Speaker 1: you say. It's starting from scratch every time you ask

835
00:49:43,040 --> 00:49:46,000
Speaker 1: it a question. Right, This is sirie Apple's personal assistant,

836
00:49:46,000 --> 00:49:47,360
Speaker 1: which is one of the first ones to hit the

837
00:49:47,400 --> 00:49:53,239
Speaker 1: market but fizzled ridiculously because of Google's UM and Amazon's

838
00:49:53,239 --> 00:49:57,000
Speaker 1: Alexa and stuff, because those actually learn and are connected

839
00:49:57,040 --> 00:50:00,200
Speaker 1: to the Internet and learn your vocal patterns and learn

840
00:50:00,280 --> 00:50:03,640
Speaker 1: your preferences and have access to that bigger network, whereas

841
00:50:03,640 --> 00:50:06,880
Speaker 1: Siri is just kind of dumb. It doesn't really do that.

842
00:50:07,120 --> 00:50:09,279
Speaker 1: They can tell you stuff about Apple, you can tell

843
00:50:09,320 --> 00:50:11,680
Speaker 1: you a lot of stuff about the history of the company,

844
00:50:12,000 --> 00:50:14,040
Speaker 1: and it can do Google searches for you. But if

845
00:50:14,080 --> 00:50:17,640
Speaker 1: you said, Kay, Sirie, what time is this movie? It

846
00:50:17,640 --> 00:50:21,239
Speaker 1: will probably give you some tangentially related to answers, but

847
00:50:21,320 --> 00:50:24,440
Speaker 1: not the exact answer you want. Time was invented shortly

848
00:50:24,480 --> 00:50:26,719
Speaker 1: after The Big Bang, as I Siri tell me a joke.

849
00:50:26,840 --> 00:50:30,480
Speaker 1: That's always fun, but no, but Google Assistance will do

850
00:50:30,560 --> 00:50:34,040
Speaker 1: that because it learns your habits and figures out what

851
00:50:34,160 --> 00:50:36,879
Speaker 1: kinds of questions you're asking. So that's what I'm saying.

852
00:50:36,880 --> 00:50:39,600
Speaker 1: When we start getting into that where we're tapped into

853
00:50:39,640 --> 00:50:43,160
Speaker 1: this larger network, it does seem like a slippery slope.

854
00:50:43,520 --> 00:50:45,880
Speaker 1: But yeah, people, even the most paranoid of my friends,

855
00:50:46,360 --> 00:50:48,239
Speaker 1: some of them are all about these home assistance and

856
00:50:48,239 --> 00:50:49,879
Speaker 1: it's this thing, this conversation we always have where it's

857
00:50:49,920 --> 00:50:53,280
Speaker 1: like I'll give it up if I can tell little

858
00:50:53,320 --> 00:50:56,040
Speaker 1: box to turn on my lights for me. Uh yeah,

859
00:50:56,320 --> 00:50:58,719
Speaker 1: I uh. I was given one of those things and

860
00:50:58,800 --> 00:51:01,760
Speaker 1: I plug it in what I'm getting and unplug it afterwards.

861
00:51:01,800 --> 00:51:04,799
Speaker 1: So it probably thinks that I am always cooking or

862
00:51:05,120 --> 00:51:08,480
Speaker 1: just don't have power at my house, which I'm fine with.

863
00:51:08,560 --> 00:51:12,719
Speaker 1: But we want to hear from you, assuming again to

864
00:51:12,800 --> 00:51:15,640
Speaker 1: the earlier question, is it really us talking to you now?

865
00:51:15,680 --> 00:51:17,280
Speaker 1: What would you like to tell us through the robot

866
00:51:17,400 --> 00:51:21,800
Speaker 1: versions of us impersonating the biological versions of us? Do

867
00:51:21,880 --> 00:51:25,880
Speaker 1: you have a home assistant? Do you use this facial

868
00:51:25,960 --> 00:51:29,920
Speaker 1: recognition software that is so often marketed as a free

869
00:51:30,400 --> 00:51:35,480
Speaker 1: recreational pursuit. Do you believe that the future will be

870
00:51:35,600 --> 00:51:40,759
Speaker 1: fraught with things that make us question digital reality? Is

871
00:51:40,800 --> 00:51:43,239
Speaker 1: there a way to combat it? Should there be? Uh?

872
00:51:43,360 --> 00:51:46,000
Speaker 1: You can find us on Instagram, you can find us

873
00:51:46,000 --> 00:51:49,040
Speaker 1: on Facebook. You can find us on Twitter. You can

874
00:51:49,080 --> 00:51:51,640
Speaker 1: find some version of us. I feel like now we

875
00:51:51,719 --> 00:51:53,799
Speaker 1: have to say you can find some version of us

876
00:51:53,840 --> 00:51:56,600
Speaker 1: on there. You can also, of course, visit our one

877
00:51:56,600 --> 00:51:59,320
Speaker 1: of our favorite places on the internet, our Facebook page.

878
00:51:59,360 --> 00:52:01,640
Speaker 1: Here's where it It's crazy, where you can see Matt

879
00:52:01,680 --> 00:52:06,640
Speaker 1: Noel and I, Um, well, what do we do there? Where?

880
00:52:06,760 --> 00:52:09,120
Speaker 1: On the Internet, and here's where it gets crazy, all

881
00:52:09,200 --> 00:52:13,680
Speaker 1: kinds of stuff. We lurk sometimes sometimes we respond and

882
00:52:13,840 --> 00:52:16,760
Speaker 1: hop around in the threads and post our own little

883
00:52:17,080 --> 00:52:19,279
Speaker 1: gifts and memes, and we have a good time in there.

884
00:52:19,320 --> 00:52:22,120
Speaker 1: It's a lot of fun. And if none of those

885
00:52:22,239 --> 00:52:25,560
Speaker 1: are quite the badger for your bag, you are in luck.

886
00:52:25,640 --> 00:52:29,280
Speaker 1: You can contact us directly. We have an email address.

887
00:52:29,440 --> 00:52:53,000
Speaker 1: It is conspiracy at how stuff works dot com.