1
00:00:04,400 --> 00:00:12,840
Speaker 1: Welcome to tech Stuff, a production from I Heart Radio. Okay, there,

2
00:00:13,039 --> 00:00:16,599
Speaker 1: I'm Lauren Vogelbaum, sitting in for Jonathan Strickland today for

3
00:00:16,760 --> 00:00:21,479
Speaker 1: I Heart Radio's International Women's Day podcast Takeover. Uh. Some

4
00:00:21,640 --> 00:00:24,239
Speaker 1: of y'all might remember me from the way back I

5
00:00:24,280 --> 00:00:26,560
Speaker 1: was Jonathan's co host here for a minute, when I

6
00:00:26,640 --> 00:00:29,480
Speaker 1: was just a ton of little baby podcaster, or from

7
00:00:29,520 --> 00:00:32,559
Speaker 1: Forward Thinking, another show that we worked on together, or

8
00:00:32,600 --> 00:00:35,680
Speaker 1: from other podcasts that I work on myself, or you

9
00:00:35,760 --> 00:00:39,400
Speaker 1: might find my voice new and strange. Um. But in

10
00:00:39,479 --> 00:00:43,520
Speaker 1: any case, hello, thank you for existing and for being

11
00:00:43,560 --> 00:00:47,440
Speaker 1: here today. In honor of International Women's Day, I wanted

12
00:00:47,440 --> 00:00:51,640
Speaker 1: to do this episode about ethics in technology, and specifically

13
00:00:51,640 --> 00:00:55,080
Speaker 1: an artificial intelligence. You know, about the ways that tech

14
00:00:55,200 --> 00:00:58,720
Speaker 1: can can hurt or help the quest to make the

15
00:00:58,760 --> 00:01:03,560
Speaker 1: world more equitable. And you might be going, aren't computers

16
00:01:03,680 --> 00:01:08,480
Speaker 1: essentially or even quintessentially unbiased? Um? You know, a program

17
00:01:08,480 --> 00:01:11,560
Speaker 1: doesn't have feelings, It only has code that it executes,

18
00:01:12,160 --> 00:01:16,039
Speaker 1: And of course that's true, but the humans who write

19
00:01:16,040 --> 00:01:20,440
Speaker 1: the code do have biases, some conscious, some unconscious, and

20
00:01:20,520 --> 00:01:23,080
Speaker 1: so the ways that we tell programs to work can

21
00:01:23,120 --> 00:01:27,000
Speaker 1: carry those biases. One example that I always think of.

22
00:01:27,200 --> 00:01:29,520
Speaker 1: And you might have seen headlines about this back in

23
00:01:29,600 --> 00:01:35,039
Speaker 1: like two about how digital cameras can behave in biased ways.

24
00:01:35,480 --> 00:01:38,920
Speaker 1: There was this whole thing where some webcams weren't tracking

25
00:01:38,959 --> 00:01:42,640
Speaker 1: and focusing on the faces of black users, and some

26
00:01:42,760 --> 00:01:47,360
Speaker 1: other cameras were flagging photos of Asian subjects because the

27
00:01:47,440 --> 00:01:51,160
Speaker 1: software insisted that their eyes were closed. In both of

28
00:01:51,200 --> 00:01:54,120
Speaker 1: these cases, it was clear that the programs had been

29
00:01:54,160 --> 00:01:58,120
Speaker 1: trained on photos of a vast majority of white faces.

30
00:01:58,440 --> 00:02:01,200
Speaker 1: The programs didn't know what to do with skin that

31
00:02:01,240 --> 00:02:05,000
Speaker 1: reflected light differently or with eyes that were a different shape.

32
00:02:05,800 --> 00:02:07,840
Speaker 1: And I will say, like this is not a purely

33
00:02:08,240 --> 00:02:10,480
Speaker 1: digital issue. UM, this sort of thing has been an

34
00:02:10,520 --> 00:02:14,320
Speaker 1: issue in photography for as long as photography has existed.

35
00:02:14,520 --> 00:02:18,760
Speaker 1: UM film stocks were originally created with only white subjects

36
00:02:18,760 --> 00:02:23,160
Speaker 1: in mind. And it wasn't until like furniture and chocolate

37
00:02:23,240 --> 00:02:26,960
Speaker 1: companies started lodging complaints with Kodak in the nineteen sixties

38
00:02:27,000 --> 00:02:29,880
Speaker 1: and seventies. The Kodak started to adjust their films to

39
00:02:29,919 --> 00:02:35,200
Speaker 1: better capture different shades of brown. But on this like

40
00:02:35,440 --> 00:02:39,000
Speaker 1: small example scale, you know, like that sucks for the

41
00:02:39,080 --> 00:02:43,520
Speaker 1: users of these cameras. But these problems are really compounded

42
00:02:43,639 --> 00:02:47,000
Speaker 1: when you start getting into big data and machine learning

43
00:02:47,120 --> 00:02:52,520
Speaker 1: and artificial intelligence. Artificial intelligence has the capacity to totally

44
00:02:52,680 --> 00:02:56,280
Speaker 1: change our world for the better. Um. Everything from making

45
00:02:56,440 --> 00:03:00,640
Speaker 1: our energy grid more efficient and more adaptable, preventing tragic

46
00:03:00,680 --> 00:03:04,920
Speaker 1: outages like we saw in Texas recently, to helping farmers

47
00:03:05,040 --> 00:03:07,640
Speaker 1: make the most of their resources and getting more fresh

48
00:03:07,680 --> 00:03:11,600
Speaker 1: foods to people who currently don't have good access to that,

49
00:03:12,280 --> 00:03:16,840
Speaker 1: to making autonomous vehicles possible, to letting your doctor just

50
00:03:16,919 --> 00:03:19,679
Speaker 1: like real quick consult every case of a disease, ever

51
00:03:19,880 --> 00:03:22,919
Speaker 1: while making a decision about how to proceed with your treatment.

52
00:03:23,560 --> 00:03:27,000
Speaker 1: Uh to I don't know, stopping the thing where you

53
00:03:27,160 --> 00:03:33,560
Speaker 1: always get served ads for the thing you just bought. Yes,

54
00:03:33,639 --> 00:03:35,760
Speaker 1: I like that T shirt, That's why I just bought

55
00:03:35,760 --> 00:03:39,400
Speaker 1: it up. Okay, Uh anyway, it's it's not It's not

56
00:03:39,480 --> 00:03:43,880
Speaker 1: as simple as um Asimov's laws of robotics when you

57
00:03:44,000 --> 00:03:47,920
Speaker 1: start getting into the wider consequences of AI UM and

58
00:03:48,240 --> 00:03:51,320
Speaker 1: you know, um, a robot may not injure a human

59
00:03:51,360 --> 00:03:54,920
Speaker 1: being or through inaction, allow a human being to come

60
00:03:54,960 --> 00:03:57,680
Speaker 1: to harm, which of course didn't even work in the

61
00:03:57,680 --> 00:04:01,480
Speaker 1: fictional world that that asthmov was set up. And robot

62
00:04:01,480 --> 00:04:04,920
Speaker 1: ethics is totally a thing that is also not simple.

63
00:04:05,040 --> 00:04:07,080
Speaker 1: And of course it is ideal if you're if your

64
00:04:07,160 --> 00:04:11,960
Speaker 1: roomba or your autonomous car does not kill you. UM,

65
00:04:11,960 --> 00:04:15,880
Speaker 1: but we're talking about designing these AI systems that will

66
00:04:16,000 --> 00:04:19,680
Speaker 1: change the way of life for whole societies. UM, it

67
00:04:19,839 --> 00:04:23,400
Speaker 1: is a big deal and it could lead to some

68
00:04:23,520 --> 00:04:28,599
Speaker 1: big problems. So we need to talk about how to

69
00:04:28,720 --> 00:04:35,000
Speaker 1: train your algorithm. Um. These big systems start small with

70
00:04:35,120 --> 00:04:39,640
Speaker 1: designers training algorithms with data sets. So right from the start, UM,

71
00:04:39,720 --> 00:04:42,040
Speaker 1: you have the issue of what data is going in

72
00:04:42,480 --> 00:04:46,279
Speaker 1: and within that data, what's being paid attention to and

73
00:04:46,320 --> 00:04:50,240
Speaker 1: what's being ignored. And now I'm not saying that these

74
00:04:50,279 --> 00:04:54,080
Speaker 1: designers are like all mustache twirling villains out there to

75
00:04:54,160 --> 00:04:56,800
Speaker 1: do evil, but they are human. You know, We're each

76
00:04:56,839 --> 00:04:59,760
Speaker 1: moving through the world with our own set of experiences.

77
00:05:00,040 --> 00:05:02,800
Speaker 1: There are so many other experiences that were bound to

78
00:05:02,960 --> 00:05:05,840
Speaker 1: fail to take some of them into consideration, or to

79
00:05:06,160 --> 00:05:10,479
Speaker 1: or to misunderstand some of those circumstances. Which is why

80
00:05:10,480 --> 00:05:12,640
Speaker 1: it's so important to have people from a variety of

81
00:05:12,680 --> 00:05:16,320
Speaker 1: backgrounds and on these projects. And right now, diversity in

82
00:05:16,360 --> 00:05:20,800
Speaker 1: tech is uh not great. UM. Back inten a whole

83
00:05:20,839 --> 00:05:23,919
Speaker 1: bunch of big tech companies got together and pledged to

84
00:05:24,240 --> 00:05:28,279
Speaker 1: increase diversity in their workforces UM and make their results public.

85
00:05:28,839 --> 00:05:32,400
Speaker 1: And every year these reports come out and the numbers

86
00:05:32,440 --> 00:05:35,880
Speaker 1: haven't changed that much in some of these categories in

87
00:05:36,600 --> 00:05:41,359
Speaker 1: six years. UH. Women are better represented now UM or

88
00:05:41,400 --> 00:05:46,360
Speaker 1: as rather having gone from around fift of the workforce

89
00:05:46,400 --> 00:05:50,960
Speaker 1: to around of the workforce at places like Google and Facebook.

90
00:05:51,680 --> 00:05:55,279
Speaker 1: But the only company that showed a comparable jump in

91
00:05:55,560 --> 00:06:00,839
Speaker 1: UM black employees was Amazon, and they're including their distribution

92
00:06:00,920 --> 00:06:05,679
Speaker 1: center employees. And other categories of underrepresented people like people

93
00:06:05,680 --> 00:06:09,279
Speaker 1: with disabilities aren't even being reported in all of these

94
00:06:09,320 --> 00:06:13,279
Speaker 1: public results UM, but studies show that they're dramatically underrepresented

95
00:06:13,279 --> 00:06:16,520
Speaker 1: in the workforce, which just isn't great. You know, when

96
00:06:16,560 --> 00:06:20,080
Speaker 1: we're designing technology like self driving cars that will need

97
00:06:20,120 --> 00:06:25,360
Speaker 1: to take into consideration the movement of wheelchair users. And

98
00:06:25,480 --> 00:06:29,840
Speaker 1: there are unfortunately instances of programs being made specifically with

99
00:06:29,960 --> 00:06:34,159
Speaker 1: bias UM, like in to sixteen when the Boston Police

100
00:06:34,160 --> 00:06:39,239
Speaker 1: Department used this social media surveillance system to flag posts

101
00:06:39,279 --> 00:06:44,440
Speaker 1: made by regular citizens who who used certain terms, for example,

102
00:06:44,920 --> 00:06:51,160
Speaker 1: colloquial Arabic Muslim words or um words like ferguson or protest,

103
00:06:51,920 --> 00:06:55,440
Speaker 1: and we are all in this, whether we like it

104
00:06:55,560 --> 00:06:58,960
Speaker 1: or not. UM Again, just as one example, everything that

105
00:06:59,000 --> 00:07:01,599
Speaker 1: we do online, from you know, what we type into

106
00:07:01,640 --> 00:07:05,880
Speaker 1: search engines and social media sites, to our location data,

107
00:07:06,040 --> 00:07:09,680
Speaker 1: to how we move our mice or like tap at

108
00:07:09,680 --> 00:07:12,760
Speaker 1: our smartphones, all of that has the potential to be

109
00:07:12,840 --> 00:07:17,440
Speaker 1: recorded and collected and sold and referenced and cross referenced

110
00:07:17,480 --> 00:07:20,360
Speaker 1: and used to track us in any number of ways.

111
00:07:21,000 --> 00:07:25,679
Speaker 1: But AI systems are a huge industry worldwide. Business spending

112
00:07:25,760 --> 00:07:30,240
Speaker 1: on artificial intelligence hit about fifty billion dollars in and

113
00:07:30,280 --> 00:07:36,960
Speaker 1: it's expected to more than double that by retail, banking, media, governments,

114
00:07:37,120 --> 00:07:41,560
Speaker 1: All kinds of industries are investing in this, and all

115
00:07:41,600 --> 00:07:45,000
Speaker 1: of this is fairly new, But of course the field

116
00:07:45,080 --> 00:07:48,560
Speaker 1: of ethics, and even computer ethics is not new at all.

117
00:07:49,480 --> 00:07:55,760
Speaker 1: So ethics and technology have always been tied together because

118
00:07:56,040 --> 00:07:59,840
Speaker 1: every time that that we humans create some new technology

119
00:07:59,840 --> 00:08:02,640
Speaker 1: that changes our world and how we interact with it

120
00:08:02,760 --> 00:08:06,720
Speaker 1: and each other, we have to reconsider our world and

121
00:08:06,880 --> 00:08:11,239
Speaker 1: our interactions. You could even argue that in that way,

122
00:08:11,400 --> 00:08:16,400
Speaker 1: like like philosophically, ethics is itself a type of technology.

123
00:08:16,640 --> 00:08:18,600
Speaker 1: But I'm not going to go that deep today. I'm

124
00:08:18,600 --> 00:08:23,120
Speaker 1: backing away from that precipice. Uh So, let's skip ahead

125
00:08:23,200 --> 00:08:26,200
Speaker 1: from you know, the beginning of human consciousness UM to

126
00:08:26,520 --> 00:08:30,800
Speaker 1: the nineteen forties, because that's when digital computers were being

127
00:08:30,840 --> 00:08:34,520
Speaker 1: invented and in the field of cybernetics got started. That's

128
00:08:34,559 --> 00:08:39,280
Speaker 1: the science of information feedback systems, right. UM. Cybernetics was

129
00:08:39,320 --> 00:08:43,160
Speaker 1: pioneered by m T mathematician Norbert Wiener and some of

130
00:08:43,200 --> 00:08:45,960
Speaker 1: his colleagues as they were working during World War Two

131
00:08:46,320 --> 00:08:49,840
Speaker 1: to develop an anti aircraft cannon that could that could

132
00:08:49,880 --> 00:08:53,720
Speaker 1: a detect and track a fast moving airplane and then

133
00:08:53,960 --> 00:08:58,640
Speaker 1: be extrapolate the airplane's probable location in the immediate future

134
00:08:58,920 --> 00:09:02,520
Speaker 1: and aim um and then see signal the firing mechanism

135
00:09:02,559 --> 00:09:08,840
Speaker 1: to fire. And that internal communication that the machine was

136
00:09:08,960 --> 00:09:13,680
Speaker 1: doing really got Weener thinking. In he published a book

137
00:09:13,720 --> 00:09:18,000
Speaker 1: called Cybernetics, or Control and Communication in the Animal and

138
00:09:18,120 --> 00:09:21,840
Speaker 1: the Machine, and in it he mused that these new

139
00:09:22,000 --> 00:09:27,000
Speaker 1: computing machines could very easily become central nervous systems for

140
00:09:27,120 --> 00:09:30,080
Speaker 1: processing all kinds of data from all kinds of instruments,

141
00:09:30,600 --> 00:09:34,360
Speaker 1: and that that potential was huge. He compared it to

142
00:09:34,440 --> 00:09:38,920
Speaker 1: nuclear weapons um. He wrote, long before Nagasaki and the

143
00:09:38,960 --> 00:09:41,920
Speaker 1: public awareness of the atomic bomb, it had occurred to

144
00:09:41,960 --> 00:09:44,679
Speaker 1: me that we were here in the presence of another

145
00:09:44,960 --> 00:09:50,040
Speaker 1: social potentiality, of unheard of importance for good and for evil.

146
00:09:51,720 --> 00:09:54,200
Speaker 1: Oh Weener expounded on this in a book he published

147
00:09:54,240 --> 00:09:58,119
Speaker 1: in nineteen fifty called The Human Use of Human Beings,

148
00:09:58,640 --> 00:10:03,600
Speaker 1: which which basically predicted that integrating computer technology into society

149
00:10:03,920 --> 00:10:07,840
Speaker 1: was going to be another revolution, just as sweeping and

150
00:10:08,000 --> 00:10:12,400
Speaker 1: messy as the Industrial Revolution, and he tried to lay

151
00:10:12,440 --> 00:10:14,600
Speaker 1: out a bit of groundwork for how to not like

152
00:10:14,920 --> 00:10:20,120
Speaker 1: totally bork it up. Spoiler alert, people borked it up anyway.

153
00:10:20,679 --> 00:10:22,840
Speaker 1: There wasn't a whole lot more work done in the

154
00:10:22,880 --> 00:10:26,320
Speaker 1: field of computer ethics until the nineteen sixties and seventies,

155
00:10:26,720 --> 00:10:31,359
Speaker 1: when computer based crime and information security and privacy concerns

156
00:10:31,559 --> 00:10:35,640
Speaker 1: had already become a problem. So by the mid sixties,

157
00:10:35,720 --> 00:10:40,480
Speaker 1: corporations and the government had both begun collecting just giant

158
00:10:40,559 --> 00:10:46,120
Speaker 1: amounts of personal data about US citizens in literally massive computers.

159
00:10:47,120 --> 00:10:50,520
Speaker 1: I guess, I guess all computers are literally massive in

160
00:10:50,520 --> 00:10:53,720
Speaker 1: that they have mass. But okay, you know what I mean. Um,

161
00:10:54,080 --> 00:10:57,280
Speaker 1: From you know, medical records to military records to legal

162
00:10:57,320 --> 00:11:01,240
Speaker 1: documents to shopping habits. And this journalist by the name

163
00:11:01,280 --> 00:11:04,959
Speaker 1: of Vance Packard wrote a book called The Naked Society

164
00:11:05,000 --> 00:11:09,600
Speaker 1: published in nineteen sixty four about the inherent privacy issue

165
00:11:09,679 --> 00:11:16,040
Speaker 1: of having that information collected and available for instantaneous reference. UH.

166
00:11:16,200 --> 00:11:20,959
Speaker 1: Something of an uproar ensued UM. Focusing as Packard's book did,

167
00:11:21,400 --> 00:11:25,880
Speaker 1: on the US government's use of citizens information UM, and

168
00:11:25,920 --> 00:11:29,000
Speaker 1: it led to just a whole bunch of data transparency

169
00:11:29,080 --> 00:11:32,520
Speaker 1: legislation over the next decade. UH. The Freedom of Information Act,

170
00:11:32,600 --> 00:11:36,080
Speaker 1: the Fair Credit Reporting Act, all designed to make sure

171
00:11:36,120 --> 00:11:39,440
Speaker 1: that citizens are able to know what data is being

172
00:11:39,480 --> 00:11:43,560
Speaker 1: collected about them by the government and how it's being used,

173
00:11:43,960 --> 00:11:47,319
Speaker 1: And to be fair, at that time, most computational power

174
00:11:47,480 --> 00:11:50,600
Speaker 1: was in the hands of the government. But this legislature

175
00:11:50,800 --> 00:11:56,120
Speaker 1: and conversation really ignored the activities of private corporations, and

176
00:11:56,160 --> 00:11:59,600
Speaker 1: it never really questioned the ethics of collecting all of

177
00:11:59,600 --> 00:12:02,679
Speaker 1: that at to in the first place. UM. And this

178
00:12:02,760 --> 00:12:05,400
Speaker 1: is apparently like a very American thing, the concept that

179
00:12:05,760 --> 00:12:09,360
Speaker 1: information is inherently good and like more is better UM.

180
00:12:09,400 --> 00:12:12,200
Speaker 1: But that is a rabbit hole for another day. But

181
00:12:12,679 --> 00:12:16,800
Speaker 1: it's not that people weren't thinking about the ethics. One

182
00:12:16,880 --> 00:12:19,240
Speaker 1: important thing that happened around the same time was that

183
00:12:19,280 --> 00:12:22,080
Speaker 1: a researcher by the name of Don Parker, who was

184
00:12:22,160 --> 00:12:26,840
Speaker 1: looking into crime being committed via computers. Parker proposed to

185
00:12:26,920 --> 00:12:32,360
Speaker 1: this leading industry professional organization, the Association for Computing Machinery,

186
00:12:32,400 --> 00:12:35,000
Speaker 1: that they develop a code of ethics for their members,

187
00:12:35,520 --> 00:12:38,079
Speaker 1: and they were like, yeah, cool, you you do that. UM.

188
00:12:38,120 --> 00:12:40,120
Speaker 1: So he headed up a committee and the a c

189
00:12:40,280 --> 00:12:43,800
Speaker 1: M adopted their first code of Ethics in ninety three.

190
00:12:44,559 --> 00:12:46,880
Speaker 1: They've updated it, I think like about once a decade

191
00:12:46,920 --> 00:12:49,640
Speaker 1: since then, UM, with the most recent update being in

192
00:12:51,160 --> 00:12:54,120
Speaker 1: It's really thoughtful. UM. One of my favorite bits from

193
00:12:54,120 --> 00:12:57,439
Speaker 1: the intro its specifies that it's quote not an algorithm

194
00:12:57,520 --> 00:13:01,000
Speaker 1: for solving ethical problems. Rather, it's as a basis for

195
00:13:01,120 --> 00:13:04,160
Speaker 1: ethical decision making. UM. You can read it if you're

196
00:13:04,160 --> 00:13:05,599
Speaker 1: into that sort of thing by going to a c

197
00:13:05,800 --> 00:13:09,439
Speaker 1: M dot org and then back to our timeline. In

198
00:13:09,520 --> 00:13:14,400
Speaker 1: nineteen seventy six, Joseph Weisenbaum, who had created the psychotherapy

199
00:13:14,440 --> 00:13:18,880
Speaker 1: mimicking chat, bought Eliza. A decade earlier, he published his

200
00:13:18,920 --> 00:13:23,240
Speaker 1: book Computer Power and Human Reason From Judgment to Calculation,

201
00:13:24,480 --> 00:13:28,280
Speaker 1: and this book was a response to the response that

202
00:13:28,400 --> 00:13:32,479
Speaker 1: he had gotten from people to his chat Bought Eliza,

203
00:13:33,040 --> 00:13:35,640
Speaker 1: and I know that Jonathan has talked about this chat

204
00:13:35,679 --> 00:13:37,960
Speaker 1: Bought on the show before. UM. It comes up a

205
00:13:38,000 --> 00:13:41,599
Speaker 1: lot and discussions about the Turing test and how convincingly

206
00:13:41,720 --> 00:13:45,480
Speaker 1: computers can approximate human communication because it was one of

207
00:13:45,480 --> 00:13:48,760
Speaker 1: the first that was effective. UM. The thing is, though,

208
00:13:48,880 --> 00:13:54,280
Speaker 1: that Weisenbaum designed Eliza as a demonstration of how bad

209
00:13:54,679 --> 00:13:59,680
Speaker 1: computers inherently are at this type of communication, but he

210
00:13:59,760 --> 00:14:03,920
Speaker 1: got the opposite response from people who tried the program

211
00:14:03,920 --> 00:14:07,280
Speaker 1: out in talking to I mean, you know, typing with

212
00:14:07,559 --> 00:14:12,720
Speaker 1: Eliza about their psychological problems. People felt like Eliza understood them,

213
00:14:12,760 --> 00:14:17,000
Speaker 1: even when they knew it was a robot. UM. So

214
00:14:17,880 --> 00:14:21,960
Speaker 1: Weisenbaum was kind of like, whoe wait uh. And so

215
00:14:22,000 --> 00:14:25,000
Speaker 1: he wrote this book to to really explain the differences

216
00:14:25,120 --> 00:14:30,560
Speaker 1: between computation and human intelligence and to assert that ethics

217
00:14:30,600 --> 00:14:34,880
Speaker 1: are imperative in the design of artificial intelligence because people

218
00:14:34,920 --> 00:14:39,440
Speaker 1: will forget the computers do not have the understanding, the wisdom,

219
00:14:39,560 --> 00:14:44,880
Speaker 1: the moral and emotional consideration of human beings. And this

220
00:14:44,960 --> 00:14:48,080
Speaker 1: is when things really started picking up in the field

221
00:14:48,080 --> 00:14:51,840
Speaker 1: of computer ethics. UM. Also in ninety this professor who

222
00:14:51,880 --> 00:14:55,560
Speaker 1: was teaching medical ethics at the time. Walter Byner noticed

223
00:14:55,680 --> 00:15:00,880
Speaker 1: how computers were complicating that field, and he became the

224
00:15:00,880 --> 00:15:04,920
Speaker 1: first person in academia to really like decide computer ethics

225
00:15:04,960 --> 00:15:08,880
Speaker 1: should be its own field of research and application UM.

226
00:15:08,880 --> 00:15:13,280
Speaker 1: He wrote that computer technology was creating whole new ethical

227
00:15:13,320 --> 00:15:17,040
Speaker 1: concerns that needed to be taken into consideration. UM, so

228
00:15:17,080 --> 00:15:20,560
Speaker 1: he popularized the term computer ethics, did speeches and workshops

229
00:15:20,560 --> 00:15:24,920
Speaker 1: and everything and jumping ahead. The first major textbook on

230
00:15:24,960 --> 00:15:28,480
Speaker 1: the subject was published called Computer Ethics, edited by one

231
00:15:28,560 --> 00:15:32,960
Speaker 1: Deborah Johnson, and Johnson disagreed with maner Um. She argued

232
00:15:33,000 --> 00:15:38,400
Speaker 1: the computers weren't creating new problems, but rather exacerbating old

233
00:15:38,480 --> 00:15:44,680
Speaker 1: problems around things like privacy, ownership, power, and responsibility. We

234
00:15:44,760 --> 00:15:47,560
Speaker 1: will get into what the future may hold after we

235
00:15:47,640 --> 00:15:58,080
Speaker 1: get back from a quick break. Welcome back. As the

236
00:15:58,160 --> 00:16:03,160
Speaker 1: computer industry group with a consumer adoption of home computing,

237
00:16:03,240 --> 00:16:06,640
Speaker 1: with with internet access growing in both the capacity and

238
00:16:06,880 --> 00:16:11,640
Speaker 1: the ubiquity of computers, just absolutely snowballing UM. The field

239
00:16:11,640 --> 00:16:15,320
Speaker 1: of computer ethics exploded UM, and more and more groups

240
00:16:15,400 --> 00:16:19,440
Speaker 1: formed to to help everyone makes sense of all of this.

241
00:16:20,040 --> 00:16:24,880
Speaker 1: For example, the Electronic Frontier Foundation was founded in there,

242
00:16:25,160 --> 00:16:29,080
Speaker 1: that advocacy and activism group that's dedicated to defending civil

243
00:16:29,160 --> 00:16:33,560
Speaker 1: liberties as technology advances. And okay, this is kind of

244
00:16:33,560 --> 00:16:36,000
Speaker 1: a side quest, but I did not notice. They actually

245
00:16:36,000 --> 00:16:39,240
Speaker 1: formed up in response to the federal seizure of a

246
00:16:39,240 --> 00:16:44,080
Speaker 1: bunch of computer equipment belonging to Steve Jackson Games, yep,

247
00:16:44,120 --> 00:16:46,960
Speaker 1: the company that brings us Apples to Apples and Munchkin

248
00:16:47,520 --> 00:16:50,160
Speaker 1: lots of other good stuff. So what happened here? What

249
00:16:50,200 --> 00:16:52,720
Speaker 1: was that was that there was this Bell South digital

250
00:16:52,760 --> 00:16:57,880
Speaker 1: document that explained how the emergency telephone system worked, and

251
00:16:57,960 --> 00:17:01,520
Speaker 1: it leaked, and the U. S. Secret Service was concerned

252
00:17:01,560 --> 00:17:04,280
Speaker 1: that having that info out there was a security concern,

253
00:17:04,359 --> 00:17:07,359
Speaker 1: you know, that hackers might overwhelm the system, um, something

254
00:17:07,400 --> 00:17:11,880
Speaker 1: like that. So so the Secret Service was conducting raids

255
00:17:12,200 --> 00:17:17,879
Speaker 1: tracking this documents digital distribution. Steve Jackson was innocent, um,

256
00:17:17,920 --> 00:17:20,520
Speaker 1: but they got this warrant, conducted this raid, took all

257
00:17:20,560 --> 00:17:24,719
Speaker 1: this stuff, um, and wound up accessing and the leading

258
00:17:24,920 --> 00:17:28,240
Speaker 1: a bunch of personal bulletin board messages from the company's

259
00:17:28,280 --> 00:17:32,439
Speaker 1: website in the process. And now the Secret Service, you know,

260
00:17:32,480 --> 00:17:35,280
Speaker 1: didn't find anything, so they gave the equipment back and

261
00:17:35,320 --> 00:17:39,600
Speaker 1: didn't press charges, but Steve Jackson was like, no, no, no,

262
00:17:40,359 --> 00:17:44,080
Speaker 1: y'all almost tanked my business. You violated the privacy of

263
00:17:44,119 --> 00:17:47,159
Speaker 1: my bulletin board users. We are pressing charges against you.

264
00:17:48,160 --> 00:17:52,439
Speaker 1: But there was really no civil rights organization that was

265
00:17:52,480 --> 00:17:55,960
Speaker 1: prepared to take on the case due to the technological

266
00:17:55,960 --> 00:17:59,159
Speaker 1: complexity of the issue. So the e f F formed

267
00:17:59,200 --> 00:18:01,960
Speaker 1: in order to bring that suit to court. Um, and

268
00:18:02,000 --> 00:18:04,480
Speaker 1: that was the first time that the court recognized the

269
00:18:04,560 --> 00:18:07,720
Speaker 1: email should have equal protection to to to to phone

270
00:18:07,720 --> 00:18:10,800
Speaker 1: calls or any other kind of communication. So so thanks,

271
00:18:10,960 --> 00:18:15,199
Speaker 1: thanks Steve Jackson for for everything. Um, the ff has

272
00:18:15,240 --> 00:18:18,239
Speaker 1: done a whole lot of important stuff. Um, they got

273
00:18:18,359 --> 00:18:22,639
Speaker 1: encryption technology taken off the list of nationally regulated weapons.

274
00:18:23,520 --> 00:18:26,840
Speaker 1: That's a separate episode though, anyway, So side quest over

275
00:18:27,000 --> 00:18:30,200
Speaker 1: back to the main quest. Um, all of this computer

276
00:18:30,240 --> 00:18:34,760
Speaker 1: stuff exploded and the field of computer ethics specialized. Um,

277
00:18:34,880 --> 00:18:39,760
Speaker 1: so now you've got internet ethics, information systems ethics, robot ethics.

278
00:18:40,720 --> 00:18:42,080
Speaker 1: And I and I do want to say, all of

279
00:18:42,119 --> 00:18:45,160
Speaker 1: this does coincide with work being done in in other

280
00:18:45,359 --> 00:18:48,640
Speaker 1: fields engine engineering and information science. Like I don't want

281
00:18:48,640 --> 00:18:52,320
Speaker 1: to imply that computer technology was the only field that

282
00:18:52,359 --> 00:18:55,640
Speaker 1: had been working with and contributing to these to these

283
00:18:55,880 --> 00:18:59,840
Speaker 1: ethical theories and the practical application of them. But like

284
00:18:59,840 --> 00:19:01,720
Speaker 1: I was saying at the top of this episode, one

285
00:19:01,720 --> 00:19:05,480
Speaker 1: of the really interesting specialties to me is in AI

286
00:19:05,560 --> 00:19:09,320
Speaker 1: ethics because it does have such potentially sweeping effects, and

287
00:19:09,440 --> 00:19:12,280
Speaker 1: also because it keeps coming up in the news for

288
00:19:12,760 --> 00:19:16,879
Speaker 1: less than rad reasons. So you have probably seen headlines

289
00:19:16,920 --> 00:19:22,119
Speaker 1: over the past few years about algorithms gone wrong. Um,

290
00:19:22,359 --> 00:19:26,639
Speaker 1: there's there's the story from where Pro Publica found that

291
00:19:26,680 --> 00:19:30,560
Speaker 1: the software used by some courts to determine the risk

292
00:19:30,680 --> 00:19:35,359
Speaker 1: of criminal defendants committing further offenses and therefore to determine

293
00:19:35,480 --> 00:19:39,840
Speaker 1: whether to detain those defendants until trial or or what

294
00:19:39,960 --> 00:19:42,840
Speaker 1: kind of bail to set for them. Um. At least

295
00:19:42,840 --> 00:19:47,359
Speaker 1: one of those algorithms, called Compass, regularly found black people

296
00:19:47,520 --> 00:19:52,480
Speaker 1: riskier and white people less risky, even when everything else

297
00:19:52,560 --> 00:19:56,640
Speaker 1: about the defendants cases were comparable. And this kind of

298
00:19:56,680 --> 00:20:00,760
Speaker 1: issue crops up in discussions about hiring software as well.

299
00:20:00,840 --> 00:20:03,880
Speaker 1: Because of a lack of care in their design, these

300
00:20:03,920 --> 00:20:08,880
Speaker 1: programs that automatically sort through resumes have ranked applicants who

301
00:20:08,920 --> 00:20:13,240
Speaker 1: have woman sounding or or black sounding names as lower

302
00:20:13,800 --> 00:20:18,920
Speaker 1: in consideration. And then there's Google's search algorithms UM from

303
00:20:20,359 --> 00:20:23,800
Speaker 1: eighteen there were all of these headlines. Reverse image searches

304
00:20:24,080 --> 00:20:28,919
Speaker 1: using photos of black people were returning images of guerrillas

305
00:20:29,680 --> 00:20:32,000
Speaker 1: because no one had taught the system how to consider

306
00:20:32,440 --> 00:20:36,399
Speaker 1: dark skin tones in people. Or if you searched the

307
00:20:36,680 --> 00:20:42,080
Speaker 1: seemingly innocuous term black girls, the first page results out

308
00:20:42,080 --> 00:20:46,600
Speaker 1: of trillions of web indexed pages included porn. Or there

309
00:20:46,640 --> 00:20:49,720
Speaker 1: was research into the search history of the white guy

310
00:20:49,880 --> 00:20:53,400
Speaker 1: who killed nine black people in a church in Charleston,

311
00:20:53,480 --> 00:20:57,920
Speaker 1: South Carolina. It's highly likely that he was radicalized, at

312
00:20:57,960 --> 00:21:01,119
Speaker 1: least in part due to the way that Google search

313
00:21:01,160 --> 00:21:05,280
Speaker 1: works UM. It's taking into account location and demographics about

314
00:21:05,359 --> 00:21:09,240
Speaker 1: him and other searchers in his area. Google search returning

315
00:21:09,440 --> 00:21:13,720
Speaker 1: white supremacist propaganda when he searched the term black on

316
00:21:13,800 --> 00:21:17,160
Speaker 1: white crime. By the way, related to this, you can

317
00:21:17,240 --> 00:21:21,320
Speaker 1: make Google give you less specialized search results. And I

318
00:21:21,359 --> 00:21:23,359
Speaker 1: am looking into that and doing it as soon as

319
00:21:23,400 --> 00:21:27,120
Speaker 1: I finished recording this this because it is continually infuriating

320
00:21:27,119 --> 00:21:30,000
Speaker 1: to me, just just when I'm doing my reading for

321
00:21:30,000 --> 00:21:32,760
Speaker 1: for podcast episodes like this and trying to find stuff

322
00:21:32,800 --> 00:21:36,120
Speaker 1: that isn't like a restaurant in my area. If I'm

323
00:21:36,119 --> 00:21:41,639
Speaker 1: talking about a larger concern anyway. UM. Also, remember how

324
00:21:41,640 --> 00:21:44,600
Speaker 1: I was talking about the flaws in programming of digital

325
00:21:44,640 --> 00:21:49,400
Speaker 1: cameras and how they sometimes have trouble discerning specific features

326
00:21:49,400 --> 00:21:52,560
Speaker 1: of people of color. UM. You know, extrapolate that out

327
00:21:52,680 --> 00:21:57,040
Speaker 1: to how facial recognition software is used with surveillance footage

328
00:21:57,160 --> 00:22:01,199
Speaker 1: by police departments. Um, the software is more likely to

329
00:22:01,240 --> 00:22:05,440
Speaker 1: make a mistake in identifying a black person's face because

330
00:22:05,440 --> 00:22:08,480
Speaker 1: the software just isn't as good as seeing that face

331
00:22:08,720 --> 00:22:11,480
Speaker 1: because of how it was programmed, which can lead to

332
00:22:11,680 --> 00:22:16,960
Speaker 1: false identifications and thus wrongful harassments and arrests. UM. Georgetown

333
00:22:17,040 --> 00:22:20,639
Speaker 1: University put out a whole report on this in called

334
00:22:20,720 --> 00:22:24,560
Speaker 1: the Perpetual Lineup. That report found that half of Americans

335
00:22:24,760 --> 00:22:28,240
Speaker 1: have photos and police facial recognition databases, by the way, UM,

336
00:22:28,280 --> 00:22:32,080
Speaker 1: which includes just lots of people with no criminal backgrounds.

337
00:22:32,400 --> 00:22:35,000
Speaker 1: And of course, even if we fix those algorithms, that

338
00:22:35,000 --> 00:22:37,560
Speaker 1: that isn't going to fix the fact that communities of

339
00:22:37,600 --> 00:22:41,399
Speaker 1: color are subject to more surveillance in the first place.

340
00:22:42,920 --> 00:22:46,600
Speaker 1: More recently, there have been headlines and a whole discussion

341
00:22:46,680 --> 00:22:50,840
Speaker 1: about this new rule that was issued in September by

342
00:22:51,000 --> 00:22:54,840
Speaker 1: the U S Department of Housing and Urban Development. And

343
00:22:54,880 --> 00:22:59,280
Speaker 1: this rule essentially makes it super difficult for banks or

344
00:22:59,400 --> 00:23:03,960
Speaker 1: landlord or homeowners insurance companies to be sued for denying

345
00:23:04,000 --> 00:23:07,840
Speaker 1: housing two people of color if an algorithm was used

346
00:23:08,000 --> 00:23:12,160
Speaker 1: to make that determination. Based on the concept that algorithms

347
00:23:12,160 --> 00:23:16,720
Speaker 1: cannot be racist, the rule was immediately challenged, UM, and

348
00:23:17,000 --> 00:23:21,679
Speaker 1: in January, President Biden issued an executive order directing the

349
00:23:21,720 --> 00:23:25,000
Speaker 1: Department of Housing and Urban Development to to examine the

350
00:23:25,080 --> 00:23:31,240
Speaker 1: rules effects. But hoof, hoof, UM, And you know it's

351
00:23:31,280 --> 00:23:35,560
Speaker 1: it's not it's not easy. None of this is easy, um.

352
00:23:35,600 --> 00:23:40,120
Speaker 1: In order to build a non biased artificial intelligence system,

353
00:23:40,280 --> 00:23:43,560
Speaker 1: we um, I mean like like humans, not like you know,

354
00:23:43,640 --> 00:23:46,840
Speaker 1: you and me, dear listener. We need to change the

355
00:23:46,880 --> 00:23:51,280
Speaker 1: systems that lead to the building of artificial intelligence. We

356
00:23:51,320 --> 00:23:56,399
Speaker 1: need to examine how the design and programming is taught, um,

357
00:23:56,440 --> 00:24:01,640
Speaker 1: how companies conduct their business, how policy is written, and

358
00:24:01,640 --> 00:24:05,119
Speaker 1: and who has access to seats at all of those tables.

359
00:24:05,520 --> 00:24:08,280
Speaker 1: I mean it's also not technically easy, Like when you

360
00:24:08,320 --> 00:24:11,520
Speaker 1: build and train these systems, just adding more diverse data

361
00:24:11,760 --> 00:24:15,800
Speaker 1: isn't magically going to make the system create better, less

362
00:24:15,840 --> 00:24:19,800
Speaker 1: biased rules. It might create conflicting rules. UM. You know

363
00:24:20,000 --> 00:24:23,320
Speaker 1: it is expensive in terms of time and effort and

364
00:24:23,320 --> 00:24:26,720
Speaker 1: and just pure physical energy to to do this work.

365
00:24:27,359 --> 00:24:31,159
Speaker 1: The pitfalls of not doing this work are tremendous. You know,

366
00:24:31,200 --> 00:24:35,720
Speaker 1: it can cause measurable hurt in people's lives. And as

367
00:24:35,760 --> 00:24:41,080
Speaker 1: one doctor Debi Chatra, material science engineer, has said, any

368
00:24:41,160 --> 00:24:47,320
Speaker 1: sufficiently advanced neglect is indistinguishable from malice. But the benefits

369
00:24:47,480 --> 00:24:53,600
Speaker 1: to doing this work are equally measurable and tremendous. One

370
00:24:53,760 --> 00:24:57,840
Speaker 1: consideration moving into the future is how to square the

371
00:24:57,960 --> 00:25:03,920
Speaker 1: very concept of ethics with an increasingly multicultural digital world. Um.

372
00:25:03,960 --> 00:25:06,160
Speaker 1: You know, not not everyone on the planet grew up

373
00:25:06,160 --> 00:25:09,440
Speaker 1: with European philosophy, going back to the ancient Greek as

374
00:25:09,440 --> 00:25:13,439
Speaker 1: the basis of their ethical conception. And we also have

375
00:25:13,520 --> 00:25:17,440
Speaker 1: to acknowledge that, um, you know, whatever a culture's philosophical basis,

376
00:25:18,040 --> 00:25:21,240
Speaker 1: it's probably rooted in some biases of its own. Um.

377
00:25:21,280 --> 00:25:24,320
Speaker 1: Just for example, just throwing it out there, if your

378
00:25:24,320 --> 00:25:29,720
Speaker 1: society has coded emotions as feminine and feminine as bad, um,

379
00:25:29,840 --> 00:25:34,040
Speaker 1: then you're probably not giving emotional harm as much weight,

380
00:25:34,119 --> 00:25:38,680
Speaker 1: if any weight, as physical harm in your considerations of justice. Um.

381
00:25:38,720 --> 00:25:41,800
Speaker 1: And you can see the effects of this in things

382
00:25:41,880 --> 00:25:44,720
Speaker 1: like the care that we give our veterans with physical

383
00:25:44,760 --> 00:25:49,600
Speaker 1: injury versus veterans with PTSD UM or just the general

384
00:25:49,600 --> 00:25:53,240
Speaker 1: ways that our society handles mental health versus physical health,

385
00:25:53,520 --> 00:25:57,920
Speaker 1: or or any kind of neurodivergence. All of this work

386
00:25:58,080 --> 00:26:02,719
Speaker 1: in artificial intelligence and the ethics thereof is really requiring

387
00:26:02,800 --> 00:26:08,080
Speaker 1: us to redefine intelligence, to fully consider what we mean

388
00:26:08,359 --> 00:26:12,359
Speaker 1: by human intelligence, what logic and emotion and experience go

389
00:26:12,440 --> 00:26:16,440
Speaker 1: into that, and the ways in which machine intelligence might differ.

390
00:26:16,880 --> 00:26:22,480
Speaker 1: The Stanford Encyclopedia Philosophy references Minsky's book The Society of

391
00:26:22,480 --> 00:26:26,760
Speaker 1: the Mind, saying we do not wish to restrict intelligence

392
00:26:27,000 --> 00:26:32,280
Speaker 1: to what would require intelligence if done by humans. And

393
00:26:32,280 --> 00:26:35,800
Speaker 1: and of course that's true, um A, I can do

394
00:26:35,920 --> 00:26:40,080
Speaker 1: stuff that we can't that that's arguably the whole point, um.

395
00:26:40,119 --> 00:26:42,439
Speaker 1: But it all has to be done with the best

396
00:26:42,640 --> 00:26:47,560
Speaker 1: of what human intelligence can be in mind, and I'm

397
00:26:47,560 --> 00:26:49,439
Speaker 1: just now realizing that, like I might have written a

398
00:26:49,520 --> 00:26:52,560
Speaker 1: forward thinking episode instead of a text stuff episode. But

399
00:26:52,560 --> 00:26:54,920
Speaker 1: but that's what I've got for you today. So if

400
00:26:55,000 --> 00:26:57,600
Speaker 1: you have enjoyed this episode and would like to hear

401
00:26:57,640 --> 00:27:01,280
Speaker 1: more from me, um, you can find me podcasts like

402
00:27:01,680 --> 00:27:05,200
Speaker 1: brain Stuff. It's a It's a daily short form general

403
00:27:05,240 --> 00:27:09,000
Speaker 1: science and culture show. UM or Savor which is a

404
00:27:09,160 --> 00:27:13,679
Speaker 1: food science and history show. Or American Shadows, which is

405
00:27:13,680 --> 00:27:16,440
Speaker 1: produced with Aaron Minky's company Grim and Mild UM It's

406
00:27:16,480 --> 00:27:20,040
Speaker 1: it's a show about some of the darker bits of

407
00:27:20,080 --> 00:27:25,040
Speaker 1: American history UH and ways in which even those dire

408
00:27:25,080 --> 00:27:29,320
Speaker 1: situations UM had light brought to them. I would like

409
00:27:29,400 --> 00:27:32,359
Speaker 1: to give a quick shout out to my friend Damien

410
00:27:32,359 --> 00:27:35,040
Speaker 1: Patrick Williams. He works in this field and has made

411
00:27:35,040 --> 00:27:36,879
Speaker 1: me more familiar with a lot of the concepts that

412
00:27:36,920 --> 00:27:38,959
Speaker 1: I talked about today. UM. You can find lots more

413
00:27:39,040 --> 00:27:43,520
Speaker 1: from him at a Future Worth thinking About dot com.

414
00:27:43,520 --> 00:27:47,520
Speaker 1: This podcast is produced by Tari Harrison and Severe. Thanks

415
00:27:47,520 --> 00:27:50,320
Speaker 1: to her for being so kind and accommodating UM and

416
00:27:50,440 --> 00:27:54,240
Speaker 1: helping me with this episode. The executive producer is Jonathan Strickland,

417
00:27:54,440 --> 00:27:56,560
Speaker 1: and thanks to him for trusting me with his podcast

418
00:27:56,600 --> 00:27:59,399
Speaker 1: for a day. Thanks to you for listening, and He'll

419
00:27:59,400 --> 00:28:06,880
Speaker 1: talk to you again. End really soon. Yeah. Text Stuff

420
00:28:06,920 --> 00:28:10,120
Speaker 1: is an I Heart Radio production. For more podcasts from

421
00:28:10,119 --> 00:28:13,880
Speaker 1: I Heart Radio, visit the i Heart Radio app, Apple Podcasts,

422
00:28:14,000 --> 00:28:16,000
Speaker 1: or wherever you listen to your favorite shows.