1
00:00:00,080 --> 00:00:02,960
Speaker 1: If you love iPhone, you'll love Apple Card. It's the

2
00:00:03,040 --> 00:00:06,920
Speaker 1: credit card designed for iPhone. It gives you unlimited daily

3
00:00:07,040 --> 00:00:10,000
Speaker 1: cash back that can earn four point four zero percent

4
00:00:10,160 --> 00:00:12,960
Speaker 1: annual percentage yield. When you open a high Yield savings

5
00:00:13,000 --> 00:00:16,280
Speaker 1: account through Apple Card, apply for Applecard in the wallet

6
00:00:16,280 --> 00:00:19,520
Speaker 1: app subject to credit approval. Savings is available to Apple

7
00:00:19,560 --> 00:00:22,800
Speaker 1: Card owners subject to eligibility. Apple Card and Savings by

8
00:00:22,840 --> 00:00:26,120
Speaker 1: Goldman Sachs Bank USA, Salt Lake City Branch, Member FDIC

9
00:00:26,520 --> 00:00:29,280
Speaker 1: terms and more at applecard dot Com. When you pop

10
00:00:29,320 --> 00:00:31,240
Speaker 1: a piece of cheese into your mouth, you're probably not

11
00:00:31,360 --> 00:00:34,280
Speaker 1: thinking about the environmental impact. But the people in the

12
00:00:34,360 --> 00:00:37,680
Speaker 1: dairy industry are. That's why they're working hard every day

13
00:00:37,720 --> 00:00:40,760
Speaker 1: to find new ways to reduce waste, conserve natural resources,

14
00:00:40,800 --> 00:00:44,360
Speaker 1: and drive down greenhouse gas emissions. How is US Dairy

15
00:00:44,400 --> 00:00:48,520
Speaker 1: tackling greenhouse gases? Many farms use anaerobic digesters to turn

16
00:00:48,560 --> 00:00:53,080
Speaker 1: the methane from manure into renewable energy that can power farms, towns,

17
00:00:53,120 --> 00:00:57,240
Speaker 1: and electric cars. Visit us dairy dot COM's Last Sustainability

18
00:00:57,280 --> 00:00:58,000
Speaker 1: to learn more.

19
00:00:58,640 --> 00:01:01,600
Speaker 2: Everyone loves getting good at advice and staying in the know.

20
00:01:02,320 --> 00:01:04,880
Speaker 2: There's nothing like getting a heads up on something before

21
00:01:04,920 --> 00:01:07,280
Speaker 2: you've even had time to think about whether you need

22
00:01:07,440 --> 00:01:11,920
Speaker 2: or want it. Well, Thankfully, AT and T provides personalized

23
00:01:11,959 --> 00:01:15,119
Speaker 2: recommendations and solutions so you get what's right for you.

24
00:01:15,680 --> 00:01:18,319
Speaker 2: Whether right for you means a plan that's better suited

25
00:01:18,360 --> 00:01:21,039
Speaker 2: for you and your family, or a product that makes

26
00:01:21,080 --> 00:01:24,960
Speaker 2: sense for you and your lifestyle. So relax and let

27
00:01:25,040 --> 00:01:29,440
Speaker 2: AT and T provide proactive recommendations to help empower your

28
00:01:29,480 --> 00:01:36,040
Speaker 2: best connected life.

29
00:01:40,240 --> 00:01:43,480
Speaker 3: Alexa, what's the best science podcast on air?

30
00:01:43,840 --> 00:01:46,040
Speaker 1: Hey? Are you trying to replace me with Alexa? What's

31
00:01:46,040 --> 00:01:46,640
Speaker 1: going on here?

32
00:01:47,560 --> 00:01:48,880
Speaker 3: Do you think you're replaceable?

33
00:01:49,080 --> 00:01:52,400
Speaker 1: There's no way an artificial intelligence could ever make jokes

34
00:01:52,520 --> 00:01:53,600
Speaker 1: nearly as funny as I am.

35
00:01:54,720 --> 00:01:57,520
Speaker 3: I think there's no way an artificial intelligence would laugh

36
00:01:57,560 --> 00:01:58,200
Speaker 3: at your jokes.

37
00:01:59,240 --> 00:02:01,440
Speaker 1: I'm pretty sure I could. I could program a pretty

38
00:02:01,480 --> 00:02:03,760
Speaker 1: dumb computer to lead out my jokes. It's called the

39
00:02:03,920 --> 00:02:04,560
Speaker 1: laugh track.

40
00:02:06,000 --> 00:02:09,600
Speaker 3: But that's Hey, that's a new change for AI. You know,

41
00:02:09,760 --> 00:02:13,960
Speaker 3: first chests, then go now science comedy.

42
00:02:14,680 --> 00:02:17,840
Speaker 1: That's right now, programs something that can find humor in

43
00:02:17,919 --> 00:02:19,040
Speaker 1: Daniel's ramblings.

44
00:02:37,000 --> 00:02:39,799
Speaker 3: Hi am Jorge and Don Daniel, and this is our podcast.

45
00:02:39,880 --> 00:02:42,560
Speaker 3: Daniel and Jorge explain the universe.

46
00:02:42,320 --> 00:02:45,200
Speaker 1: In which we try to download everything we know about

47
00:02:45,240 --> 00:02:49,079
Speaker 1: the universe, episode by episode into your brain, whether you're

48
00:02:49,120 --> 00:02:54,079
Speaker 1: a real person or an artificial intelligence listening to our podcast.

49
00:02:53,800 --> 00:02:55,799
Speaker 3: Well trying to sound intelligent about.

50
00:02:55,560 --> 00:02:59,320
Speaker 1: It while writing your own humor for the Open Mic

51
00:02:59,360 --> 00:03:07,160
Speaker 1: AI Night. The topic of today's podcast is what is

52
00:03:07,320 --> 00:03:08,440
Speaker 1: artificial intelligence?

53
00:03:08,639 --> 00:03:11,760
Speaker 3: And very importantly is it dangerous?

54
00:03:11,960 --> 00:03:14,800
Speaker 1: That's right? Should you be looking at your window for

55
00:03:14,840 --> 00:03:17,000
Speaker 1: the first signs of the robot revolution?

56
00:03:17,280 --> 00:03:19,080
Speaker 3: Should you be afraid of your Alexa?

57
00:03:19,120 --> 00:03:22,680
Speaker 1: Should you be worried about that robot vacuum cleaner getting

58
00:03:22,720 --> 00:03:25,280
Speaker 1: resentful for having to do all the dirty work and

59
00:03:25,400 --> 00:03:27,079
Speaker 1: eating your face off in the middle of the night.

60
00:03:27,240 --> 00:03:30,440
Speaker 3: Oh geez, that's a bit dark.

61
00:03:30,880 --> 00:03:36,400
Speaker 1: It seems kind of sinister, doesn't it. It's like sitting there, circling, circling, circling, waiting, waiting, waiting.

62
00:03:36,680 --> 00:03:39,640
Speaker 3: I think those things are creepy, right, Maybe it wants

63
00:03:39,680 --> 00:03:41,480
Speaker 3: to you clean your face.

64
00:03:41,400 --> 00:03:44,680
Speaker 1: It wants see that's the question. Does a robot vacuum

65
00:03:44,680 --> 00:03:47,800
Speaker 1: cleaner want anything? What does it mean for it to want?

66
00:03:48,080 --> 00:03:50,240
Speaker 1: What is it like to be a robot vacuum cleaner?

67
00:03:50,600 --> 00:03:52,400
Speaker 1: The next great paper in philosophy?

68
00:03:52,600 --> 00:03:54,840
Speaker 3: So this is kind of in the zeitgeist right now.

69
00:03:54,880 --> 00:03:58,000
Speaker 3: I mean, people are really excited about artificial intelligence. But

70
00:03:58,040 --> 00:04:00,400
Speaker 3: at the same time there are big names like Elon

71
00:04:00,480 --> 00:04:04,600
Speaker 3: Musk kind of warning people like, hey, artificial intelligence not

72
00:04:04,720 --> 00:04:05,440
Speaker 3: such a good idea.

73
00:04:05,640 --> 00:04:07,360
Speaker 1: That's right, it's a huge topic. I mean, you drive

74
00:04:07,400 --> 00:04:10,560
Speaker 1: around like San Francisco, you see artificial intelligence, machine learning,

75
00:04:10,640 --> 00:04:13,200
Speaker 1: deep learning. It's on billboards. Even you know you want

76
00:04:13,200 --> 00:04:14,760
Speaker 1: to get a million bucks for your new company, you

77
00:04:14,800 --> 00:04:17,680
Speaker 1: just say the words AI, deep learning, and boom, people

78
00:04:17,720 --> 00:04:21,240
Speaker 1: are throwing cash at you. Right, deep learning, in the

79
00:04:21,279 --> 00:04:25,320
Speaker 1: deepest learning. It's definitely part of the cultural moment. And

80
00:04:25,720 --> 00:04:27,640
Speaker 1: you see that reflected not just in like what deep

81
00:04:27,640 --> 00:04:29,760
Speaker 1: thinkers are saying, but also in like science fiction. You know,

82
00:04:29,800 --> 00:04:32,680
Speaker 1: a lot of the near term dystopian these days is

83
00:04:32,720 --> 00:04:36,080
Speaker 1: about how AI will take over and the dangers of AI.

84
00:04:36,560 --> 00:04:39,120
Speaker 1: Another way, like thirty years ago is about the dangers

85
00:04:39,120 --> 00:04:43,039
Speaker 1: of radiation. Right, that was the new dangerous thing physicist

86
00:04:43,040 --> 00:04:45,839
Speaker 1: that invented. Now the new dangerous technology that we're all

87
00:04:45,839 --> 00:04:46,680
Speaker 1: worried about is AI.

88
00:04:46,960 --> 00:04:48,440
Speaker 3: It's a new promise in peril.

89
00:04:49,920 --> 00:04:52,159
Speaker 1: But yeah, AI. Every piece of technology is a double

90
00:04:52,240 --> 00:04:53,880
Speaker 1: edged sword, right, you can use it for good. You

91
00:04:53,880 --> 00:04:56,560
Speaker 1: can use it for evil. But AI is special because

92
00:04:56,560 --> 00:04:59,000
Speaker 1: it's not just technology. It's not just a tool that

93
00:04:59,040 --> 00:05:03,600
Speaker 1: people use trully that has independence, that has autonomy. And

94
00:05:03,600 --> 00:05:05,400
Speaker 1: that's why it's such a vexing question.

95
00:05:05,680 --> 00:05:08,760
Speaker 3: Well people, I'm sure people everyone associates it with robots

96
00:05:08,800 --> 00:05:11,120
Speaker 3: and machines and computers. But we were kind of wondering

97
00:05:11,200 --> 00:05:15,280
Speaker 3: if people actually knew what artificial intelligence was, like, what

98
00:05:15,360 --> 00:05:18,680
Speaker 3: makes it work, what makes it different than real intelligence.

99
00:05:19,440 --> 00:05:21,120
Speaker 1: I bet you that ninety five percent of the people

100
00:05:21,160 --> 00:05:24,040
Speaker 1: who say the phrase artificial intelligence don't actually know what

101
00:05:24,080 --> 00:05:27,920
Speaker 1: they're talking about, which is probably true for most topics.

102
00:05:28,279 --> 00:05:32,760
Speaker 3: Technology, it's true for me probably, I'm sure. But we're

103
00:05:32,800 --> 00:05:36,320
Speaker 3: wondering if you guys out there knew what artificial intelligence was.

104
00:05:36,680 --> 00:05:39,119
Speaker 3: And so, as usual, Daniel went out and asked people

105
00:05:39,120 --> 00:05:39,400
Speaker 3: in the.

106
00:05:39,320 --> 00:05:41,559
Speaker 1: Street, and here's what they had to say.

107
00:05:43,040 --> 00:05:45,560
Speaker 4: It's this idea that we can create some type of

108
00:05:45,800 --> 00:05:49,800
Speaker 4: material thing that could think on its own ultimately, and do.

109
00:05:49,720 --> 00:05:51,720
Speaker 1: You think it's something we should be concerned about? Is

110
00:05:51,760 --> 00:05:53,160
Speaker 1: it ever going to be a threat to humanity?

111
00:05:53,839 --> 00:05:56,960
Speaker 4: I mean possibly, but I mean we don't know everything.

112
00:05:56,960 --> 00:05:59,279
Speaker 4: We can know the bounds of what could be a threat,

113
00:05:59,320 --> 00:06:00,000
Speaker 4: what could not be a threat.

114
00:06:00,640 --> 00:06:04,160
Speaker 5: Yeah, it's AI, and it's just the stuff that's used

115
00:06:04,160 --> 00:06:07,480
Speaker 5: in various technological applications. Basically, they're just kind of like

116
00:06:08,320 --> 00:06:12,320
Speaker 5: trying to make machines replicate certain aspects of human intelligence.

117
00:06:12,680 --> 00:06:13,279
Speaker 5: Stuff like that.

118
00:06:13,440 --> 00:06:14,839
Speaker 1: Okay, And do you think it could ever be a

119
00:06:14,880 --> 00:06:16,799
Speaker 1: threat to humanity? Is something we should be worried about?

120
00:06:17,160 --> 00:06:19,960
Speaker 5: I guess since I don't have a particularly strong opinion

121
00:06:20,040 --> 00:06:23,440
Speaker 5: on it, I don't think so, So I guess I'll

122
00:06:23,440 --> 00:06:24,160
Speaker 5: say no for now.

123
00:06:25,000 --> 00:06:29,240
Speaker 4: I'm assuming that's the idea that computers and electrons can

124
00:06:29,760 --> 00:06:30,840
Speaker 4: have like sentience.

125
00:06:31,640 --> 00:06:34,839
Speaker 1: Right, Are you worried that computers would would a day

126
00:06:34,880 --> 00:06:36,440
Speaker 1: take over and make us their slaves?

127
00:06:36,600 --> 00:06:37,159
Speaker 6: And not?

128
00:06:37,400 --> 00:06:40,120
Speaker 1: Really, I don't think it will come to that point.

129
00:06:40,279 --> 00:06:42,880
Speaker 3: All Right, those are pretty sophisticated answers. I like the

130
00:06:42,920 --> 00:06:46,520
Speaker 3: ones that said, oh, artificial intelligence, that's just AI, right,

131
00:06:47,000 --> 00:06:47,880
Speaker 3: Like that's an answer.

132
00:06:49,880 --> 00:06:52,320
Speaker 1: So that's an answer to every question. You know, what

133
00:06:52,480 --> 00:07:01,920
Speaker 1: is Google blocks zabie brom, Oh, that's just Gahcronyms.

134
00:06:58,440 --> 00:07:02,240
Speaker 3: Can make you look intelligent, that's the real artificial intelligence.

135
00:07:02,360 --> 00:07:08,279
Speaker 1: Just speak an acronym, acronym intelligence you No, but people

136
00:07:08,320 --> 00:07:10,560
Speaker 1: had some sense that it's you know, something that can

137
00:07:10,600 --> 00:07:13,600
Speaker 1: think for itself, or something you can do something for you,

138
00:07:13,840 --> 00:07:16,800
Speaker 1: or create something that can think by itself. There's definitely

139
00:07:16,920 --> 00:07:18,560
Speaker 1: the negative idea is definitely out there.

140
00:07:19,000 --> 00:07:21,200
Speaker 3: They use it in relation to what it can do.

141
00:07:21,520 --> 00:07:25,600
Speaker 1: That's right, Yeah, exactly what what what's the new capability

142
00:07:25,640 --> 00:07:28,480
Speaker 1: that defines it? Yeah? Right, yeah, And that's It's a

143
00:07:28,480 --> 00:07:30,880
Speaker 1: fascinating way think about it, you know. And it's definitely

144
00:07:30,880 --> 00:07:31,600
Speaker 1: a tricky.

145
00:07:31,360 --> 00:07:33,400
Speaker 3: Question, right because I guess we know it in the

146
00:07:33,440 --> 00:07:36,680
Speaker 3: context of using them for things, right, Like, we're people

147
00:07:36,680 --> 00:07:38,840
Speaker 3: going to just create AI because we want to create

148
00:07:38,840 --> 00:07:41,080
Speaker 3: our official beings. It's like, so it can help us.

149
00:07:41,440 --> 00:07:44,600
Speaker 1: I want to create artificial beings. What's what's wrong with that?

150
00:07:44,600 --> 00:07:48,000
Speaker 1: That sounds pretty awesome. Create a whole army of physics

151
00:07:48,280 --> 00:07:50,680
Speaker 1: artificial physics grad students. That sounds pretty cleol. Do your

152
00:07:50,760 --> 00:07:53,840
Speaker 1: kids know well, yeah, you mean, are they worried about

153
00:07:53,920 --> 00:07:55,400
Speaker 1: competing with my digital children?

154
00:07:55,920 --> 00:07:58,800
Speaker 3: My natural They know you'd rather have artificial children.

155
00:07:59,360 --> 00:08:02,240
Speaker 1: I didn't say I'd rather have artificial gender in addition

156
00:08:02,560 --> 00:08:05,040
Speaker 1: to my beautiful, wonderful natural children, which I should not

157
00:08:05,040 --> 00:08:07,200
Speaker 1: be talking about on this podcast. I'd love to have

158
00:08:07,240 --> 00:08:10,600
Speaker 1: a whole you know, cadre of artificial children to do.

159
00:08:10,560 --> 00:08:13,680
Speaker 3: My bidding, unlike your real children who won't do your bidding.

160
00:08:14,720 --> 00:08:20,040
Speaker 1: So it's somebody listening children And that sort of goes

161
00:08:20,040 --> 00:08:22,200
Speaker 1: to the heart of the question. You know, if you

162
00:08:22,320 --> 00:08:25,400
Speaker 1: created a digital being with artificial intelligence, would it listen

163
00:08:25,440 --> 00:08:27,720
Speaker 1: to you or would it make its own decisions? Right?

164
00:08:27,800 --> 00:08:29,680
Speaker 1: And so that's what we thought it would be interesting

165
00:08:29,720 --> 00:08:32,040
Speaker 1: to dig into, like what is artificial intelligence?

166
00:08:32,080 --> 00:08:33,640
Speaker 3: If it just did what you told it to do,

167
00:08:33,800 --> 00:08:36,640
Speaker 3: it wouldn't maybe be in an artificial intelligence.

168
00:08:36,280 --> 00:08:38,120
Speaker 1: You're saying nobody smart should listen to you. Is that

169
00:08:38,120 --> 00:08:38,520
Speaker 1: what you're saying.

170
00:08:40,120 --> 00:08:44,680
Speaker 3: I'm saying they should decide for themselves whether I'm worth following. Yeah,

171
00:08:48,120 --> 00:08:50,560
Speaker 3: so let's break it down for people. Daniel, what is

172
00:08:50,800 --> 00:08:52,240
Speaker 3: artificial intelligence?

173
00:08:52,360 --> 00:08:54,360
Speaker 1: Well, you should listen to this podcast and that'll give

174
00:08:54,400 --> 00:08:54,880
Speaker 1: you the answer.

175
00:08:55,040 --> 00:08:56,079
Speaker 3: Done.

176
00:08:57,880 --> 00:09:00,000
Speaker 1: Well, you know, I think to understand what artificial intelligen

177
00:09:00,080 --> 00:09:01,480
Speaker 1: it is, we should think for a moment about what

178
00:09:01,520 --> 00:09:02,679
Speaker 1: do we mean by intelligence?

179
00:09:02,760 --> 00:09:02,880
Speaker 3: Right?

180
00:09:02,960 --> 00:09:05,520
Speaker 1: And very simply, intelligence is just the ability to learn,

181
00:09:05,920 --> 00:09:08,320
Speaker 1: is to find patterns to extrapolate from them.

182
00:09:08,440 --> 00:09:11,280
Speaker 3: Really, that's how you but like a dog can learn,

183
00:09:11,480 --> 00:09:13,319
Speaker 3: but a dog you wouldn't say it's intelligent.

184
00:09:13,360 --> 00:09:15,920
Speaker 1: Would you absolutely? I would say a dog is intelligent.

185
00:09:16,160 --> 00:09:17,920
Speaker 1: You can teach a dog, you can train a dog.

186
00:09:17,960 --> 00:09:20,760
Speaker 3: It's more intelligent than a rock. But would you say

187
00:09:20,840 --> 00:09:22,840
Speaker 3: a lot, Yeah, buy a lot.

188
00:09:23,520 --> 00:09:25,560
Speaker 1: Oh my gosh, if you like, never interact with a dog.

189
00:09:25,600 --> 00:09:28,240
Speaker 1: A dog is like a living sentient being. It feels,

190
00:09:28,280 --> 00:09:31,800
Speaker 1: it experiences, It definitely learns. It can recognize you. Yeah,

191
00:09:31,840 --> 00:09:33,719
Speaker 1: I mean dogs can do complicated things. A dog is

192
00:09:33,720 --> 00:09:34,840
Speaker 1: a perfect example. Second.

193
00:09:35,000 --> 00:09:36,880
Speaker 3: You know, I wouldn't trust it to do my taxes.

194
00:09:36,960 --> 00:09:39,080
Speaker 3: You know, well, I don't know.

195
00:09:39,080 --> 00:09:41,000
Speaker 1: Compared to our tax accountant, it might do a pretty

196
00:09:41,000 --> 00:09:41,320
Speaker 1: good job.

197
00:09:41,360 --> 00:09:43,360
Speaker 3: I mean, you could say that's an intelligent dog, but

198
00:09:43,440 --> 00:09:46,000
Speaker 3: you wouldn't say, like, that's the epitome of intelligence.

199
00:09:46,200 --> 00:09:48,800
Speaker 1: I wouldn't say the dogs are the most intelligent beings

200
00:09:48,800 --> 00:09:50,760
Speaker 1: in the universe. But that's not what we're talking about. We're

201
00:09:50,760 --> 00:09:54,360
Speaker 1: talking about do they have intelligence an example, because they

202
00:09:54,360 --> 00:09:56,480
Speaker 1: can learn, you can train them, and you The cool

203
00:09:56,520 --> 00:09:58,600
Speaker 1: thing about an intelligent being is that you can train

204
00:09:58,679 --> 00:10:01,160
Speaker 1: it to do something, and if you don't know how

205
00:10:01,200 --> 00:10:03,679
Speaker 1: to do it, say, for example, you want your dog

206
00:10:04,240 --> 00:10:07,680
Speaker 1: to recognize you, right, but tear the face off anybody

207
00:10:07,679 --> 00:10:10,959
Speaker 1: who tries to break into the house, right, guard dog? Okay,

208
00:10:11,240 --> 00:10:13,360
Speaker 1: So you can train a dog. You reward it when

209
00:10:13,400 --> 00:10:15,319
Speaker 1: it does the right thing, and you punish it when

210
00:10:15,320 --> 00:10:17,839
Speaker 1: it does the wrong thing. You don't know how to

211
00:10:17,880 --> 00:10:20,839
Speaker 1: like build a being that does that, that like recognizes

212
00:10:20,880 --> 00:10:23,840
Speaker 1: your face and recognizes strangers faces and makes these decisions.

213
00:10:23,920 --> 00:10:26,719
Speaker 1: That's a hard task, you know, it's not easy to do.

214
00:10:27,080 --> 00:10:29,280
Speaker 1: But you can train a dog. A dog can learn

215
00:10:29,440 --> 00:10:31,559
Speaker 1: how to solve this problem, and all you need to

216
00:10:31,600 --> 00:10:33,640
Speaker 1: do to train it is to reward it and punish it.

217
00:10:33,720 --> 00:10:36,040
Speaker 3: So you're saying, just the ability to sort of learn

218
00:10:36,120 --> 00:10:40,080
Speaker 3: from your mistakes or learn from your surroundings. That's what

219
00:10:40,080 --> 00:10:41,160
Speaker 3: you would call intelligence.

220
00:10:41,400 --> 00:10:43,760
Speaker 1: Yeah, and dogs have less of it than we do,

221
00:10:44,160 --> 00:10:46,800
Speaker 1: you know, and more of it than cats and mice.

222
00:10:47,440 --> 00:10:49,640
Speaker 1: But they have some of it for sure, okay, which

223
00:10:49,679 --> 00:10:52,120
Speaker 1: is what makes them trainable. And you know, I wonder sometimes,

224
00:10:52,120 --> 00:10:55,360
Speaker 1: because dogs can be trained, right, nobody ever trains their cat.

225
00:10:55,559 --> 00:10:59,040
Speaker 1: What does that say about a cat's intelligence. I've always thought,

226
00:10:59,120 --> 00:11:01,240
Speaker 1: I mean, I love cat, but I've always thought dogs

227
00:11:01,240 --> 00:11:03,959
Speaker 1: are probably smarter than cats because you can train.

228
00:11:03,840 --> 00:11:07,000
Speaker 3: Them, right, or maybe cats are more intelligent in that

229
00:11:07,040 --> 00:11:10,320
Speaker 3: they're not they don't allow themselves to be trained by humans.

230
00:11:10,880 --> 00:11:13,360
Speaker 1: Right, And rocks, by that metric, are the most intelligent

231
00:11:13,400 --> 00:11:15,840
Speaker 1: because they completely ignore you. Right, You see the fallacy

232
00:11:15,840 --> 00:11:16,800
Speaker 1: of that argument right there.

233
00:11:16,920 --> 00:11:19,640
Speaker 3: But I mean maybe there's sort of sort of like

234
00:11:19,640 --> 00:11:22,640
Speaker 3: a hump, right, like, as you get more intelligent, you're

235
00:11:22,720 --> 00:11:26,160
Speaker 3: easily more trainable, trainable, but somebody you get so intelligent

236
00:11:26,280 --> 00:11:28,640
Speaker 3: that you rebel against your masters.

237
00:11:28,880 --> 00:11:30,800
Speaker 1: And so how do you tell the difference between something

238
00:11:30,840 --> 00:11:34,480
Speaker 1: that's totally unintelligent and something that's so intelligent it completely

239
00:11:34,520 --> 00:11:35,000
Speaker 1: ignores you.

240
00:11:35,320 --> 00:11:37,600
Speaker 3: Yeah, I don't know, deep.

241
00:11:37,480 --> 00:11:41,200
Speaker 1: Question or believes all the rocks are probably thinking about him.

242
00:11:41,280 --> 00:11:43,600
Speaker 3: Well sure, I mean if you use the ability to

243
00:11:43,800 --> 00:11:46,520
Speaker 3: listen to what I say as a benchmark of intelligence,

244
00:11:46,559 --> 00:11:49,719
Speaker 3: and then yeah, there's something super intelligent could be just

245
00:11:49,760 --> 00:11:51,560
Speaker 3: as smart as the rock. But obviously a cat is

246
00:11:51,559 --> 00:11:54,920
Speaker 3: still making decisions and acting and you know, doing things,

247
00:11:55,440 --> 00:11:57,800
Speaker 3: so it's intelligent. But maybe it's much more intelligent than

248
00:11:57,800 --> 00:12:00,360
Speaker 3: a dog because it chooses not to listen to us.

249
00:12:01,040 --> 00:12:02,280
Speaker 1: All right, I think we need to have a whole

250
00:12:02,280 --> 00:12:05,760
Speaker 1: other podcast on who smarter counts or does And before

251
00:12:05,840 --> 00:12:08,800
Speaker 1: we do that, we will collect some data to answer

252
00:12:08,800 --> 00:12:11,760
Speaker 1: this question. But I think with the question we were

253
00:12:11,760 --> 00:12:15,160
Speaker 1: focusing on is what is artificial intelligence? So natural intelligence

254
00:12:15,240 --> 00:12:18,640
Speaker 1: just the ability of an animal to learn. Artificial intelligence

255
00:12:18,640 --> 00:12:22,079
Speaker 1: would be if something artificial that we create has that

256
00:12:22,120 --> 00:12:23,280
Speaker 1: same property.

257
00:12:23,520 --> 00:12:27,679
Speaker 3: The ability to change the way it processes things in

258
00:12:27,720 --> 00:12:30,360
Speaker 3: response to what it sees about the world.

259
00:12:30,480 --> 00:12:34,000
Speaker 1: Yeah, artificial intelligence is a very broad field with lots

260
00:12:34,040 --> 00:12:36,640
Speaker 1: of elements that we couldn't cover in just one episode

261
00:12:36,679 --> 00:12:39,240
Speaker 1: of a podcast. But let's just talk today about one

262
00:12:39,360 --> 00:12:43,640
Speaker 1: important subfield of AI, which is machine learning, or more specifically,

263
00:12:43,640 --> 00:12:46,240
Speaker 1: I would say, let's focus on training, right, can you

264
00:12:46,360 --> 00:12:50,680
Speaker 1: build something and something artificial that can be trained? Right?

265
00:12:51,280 --> 00:12:54,160
Speaker 1: And I think I think let's talk for a moment about,

266
00:12:54,320 --> 00:12:56,800
Speaker 1: you know, how how normal computers work, and then we

267
00:12:56,800 --> 00:12:59,959
Speaker 1: can talk about how computers, smart computers, computers that can

268
00:13:00,120 --> 00:13:02,360
Speaker 1: or in computers with artificial intelligence, how they work.

269
00:13:02,520 --> 00:13:04,240
Speaker 3: I think I think we should keep talking about cats

270
00:13:04,240 --> 00:13:05,120
Speaker 3: and dogs.

271
00:13:06,200 --> 00:13:08,240
Speaker 1: All right, we'll talk about cats and dogs, but first

272
00:13:08,600 --> 00:13:15,800
Speaker 1: let's take a quick break. With big wireless providers, what

273
00:13:15,840 --> 00:13:18,600
Speaker 1: you see is never what you get. Somewhere between the

274
00:13:18,600 --> 00:13:20,760
Speaker 1: store and your first month's bill. The price you thoughts

275
00:13:20,800 --> 00:13:24,440
Speaker 1: you were paying magically skyrockets. With mint Mobile, You'll never

276
00:13:24,520 --> 00:13:27,360
Speaker 1: have to worry about Gotcha's ever again. When mint Mobile

277
00:13:27,400 --> 00:13:29,640
Speaker 1: says fifteen dollars a month for a three month plan,

278
00:13:29,920 --> 00:13:32,240
Speaker 1: they really mean it. I've used mint Mobile and the

279
00:13:32,240 --> 00:13:35,000
Speaker 1: call quality is always so crisp and so clear. I

280
00:13:35,040 --> 00:13:37,200
Speaker 1: can recommend it to you. So say bye bye to

281
00:13:37,200 --> 00:13:41,720
Speaker 1: your overpriced wireless plans, jaw dropping monthly bills and unexpected overages.

282
00:13:41,800 --> 00:13:44,199
Speaker 1: You can use your own phone with any mint Mobile

283
00:13:44,200 --> 00:13:47,600
Speaker 1: plan and bring your phone number along with your existing contacts.

284
00:13:47,720 --> 00:13:50,080
Speaker 1: So dit your overpriced wireless with mint Mobiles deal and

285
00:13:50,080 --> 00:13:53,160
Speaker 1: get three months a premium wireless service for fifteen bucks

286
00:13:53,200 --> 00:13:55,000
Speaker 1: a month. To get this new customer offer and your

287
00:13:55,040 --> 00:13:57,480
Speaker 1: new three month premium wireless plan for just fifteen bucks

288
00:13:57,480 --> 00:14:00,199
Speaker 1: a month, go to mintmobile dot com slash universe. It's

289
00:14:00,280 --> 00:14:02,959
Speaker 1: mintmobile dot com slash universe. Cut your wireless build to

290
00:14:03,000 --> 00:14:05,800
Speaker 1: fifteen bucks a month. At mintmobile dot com slash universe,

291
00:14:05,840 --> 00:14:08,680
Speaker 1: forty five dollars upfront payment required equivalent to fifteen dollars

292
00:14:08,720 --> 00:14:11,080
Speaker 1: per month new customers on first three month plan only.

293
00:14:11,200 --> 00:14:14,160
Speaker 1: Speeds slower about forty gigabytes on unlimited plan. Additional taxi

294
00:14:14,160 --> 00:14:16,760
Speaker 1: speeds and restrictions apply. See mint mobile for details.

295
00:14:16,920 --> 00:14:20,000
Speaker 7: AI might be the most important new computer technology ever.

296
00:14:20,240 --> 00:14:23,440
Speaker 7: It's storming every industry and literally billions of dollars are

297
00:14:23,480 --> 00:14:27,720
Speaker 7: being invested, so buckle up. The problem is that AI

298
00:14:27,840 --> 00:14:30,640
Speaker 7: needs a lot of speed and processing power, So how

299
00:14:30,680 --> 00:14:33,840
Speaker 7: do you compete without cost spiraling out of control. It's

300
00:14:33,840 --> 00:14:36,160
Speaker 7: time to upgrade to the next generation of the cloud.

301
00:14:36,520 --> 00:14:41,680
Speaker 7: Oracle Cloud Infrastructure or OCI. OCI is a single platform

302
00:14:41,720 --> 00:14:46,520
Speaker 7: for your infrastructure, database, application development, and AI needs. OCI

303
00:14:46,720 --> 00:14:49,840
Speaker 7: has forty eight times the bandwidth of other clouds, offers

304
00:14:49,880 --> 00:14:53,200
Speaker 7: one consistent price instead of variable regional pricing, and of

305
00:14:53,240 --> 00:14:56,320
Speaker 7: course nobody does data better than Oracle. So now you

306
00:14:56,360 --> 00:14:58,720
Speaker 7: can train your AI models at twice the speed and

307
00:14:58,840 --> 00:15:01,000
Speaker 7: less than half the cost of a other clouds. If

308
00:15:01,040 --> 00:15:03,240
Speaker 7: you want to do more and spend less, like Uber

309
00:15:03,360 --> 00:15:06,000
Speaker 7: eight by eight and Data Bricks Mosaic, take a free

310
00:15:06,040 --> 00:15:09,960
Speaker 7: test drive of OCI at Oracle dot com slash strategic.

311
00:15:10,280 --> 00:15:15,520
Speaker 7: That's Oracle dot com slash Strategic Oracle dot com slash Strategic.

312
00:15:15,800 --> 00:15:18,680
Speaker 1: If you love iPhone, you'll love Apple Card. It's the

313
00:15:18,760 --> 00:15:22,640
Speaker 1: credit card designed for iPhone. It gives you unlimited daily

314
00:15:22,760 --> 00:15:25,720
Speaker 1: cash back that can earn four point four zero percent

315
00:15:25,880 --> 00:15:28,680
Speaker 1: annual percentage yield. When you open a high yield savings

316
00:15:28,720 --> 00:15:32,280
Speaker 1: account through Applecard. Apply for Applecard in the wallet app,

317
00:15:32,400 --> 00:15:35,920
Speaker 1: subject to credit approval. Savings is available to Applecard owners

318
00:15:35,920 --> 00:15:39,160
Speaker 1: subject to eligibility. Apple Card and Savings by Goldman Sachs

319
00:15:39,240 --> 00:15:42,640
Speaker 1: Bank USA Salt Lake City Branch Member FDIC, terms and

320
00:15:42,720 --> 00:15:55,160
Speaker 1: more at applecard dot com. Let's talk about what computers

321
00:15:55,200 --> 00:15:55,600
Speaker 1: can do.

322
00:15:55,880 --> 00:16:00,320
Speaker 3: Yeah, let's cause computers are smart. Right. You can a

323
00:16:00,320 --> 00:16:03,400
Speaker 3: computer to do smart things, but that doesn't necessarily mean

324
00:16:03,400 --> 00:16:04,440
Speaker 3: it has intelligence.

325
00:16:04,680 --> 00:16:06,920
Speaker 1: That's right. There's a difference between a computer that can

326
00:16:06,960 --> 00:16:09,680
Speaker 1: do something and a computer that can learn something. Right.

327
00:16:09,920 --> 00:16:12,840
Speaker 1: The way I think about non intelligent computers is the

328
00:16:12,840 --> 00:16:15,360
Speaker 1: way you sort of think about machines. Right. You can

329
00:16:15,760 --> 00:16:18,560
Speaker 1: tell them what to do, and they do exactly what

330
00:16:18,640 --> 00:16:21,120
Speaker 1: you tell them, regardless of whether it's the right thing.

331
00:16:21,280 --> 00:16:23,240
Speaker 1: You don't give them like a goal and say, hey,

332
00:16:23,280 --> 00:16:25,680
Speaker 1: I just want the house to be clean. Figure it out.

333
00:16:25,920 --> 00:16:27,720
Speaker 1: You have to tell them exactly what to do. You say,

334
00:16:27,920 --> 00:16:31,320
Speaker 1: step over here, move the broom this way, step over there,

335
00:16:31,480 --> 00:16:33,080
Speaker 1: you know, and if it's not cleaning the house because

336
00:16:33,080 --> 00:16:35,240
Speaker 1: they're stuck in a corner or they're you know, fell

337
00:16:35,280 --> 00:16:37,560
Speaker 1: on on their butts or whatever, they don't care. They

338
00:16:37,600 --> 00:16:39,960
Speaker 1: just tell it do exactly what you tell them to do.

339
00:16:40,080 --> 00:16:42,520
Speaker 1: They have no sort of larger sense of what's important.

340
00:16:42,680 --> 00:16:46,840
Speaker 3: It just follows instructions, just follows the recipe you gave it.

341
00:16:47,040 --> 00:16:49,160
Speaker 1: That's right. It's like a wind up toy, you know,

342
00:16:49,600 --> 00:16:51,280
Speaker 1: you wind it up, you give it some energy, and

343
00:16:51,320 --> 00:16:53,400
Speaker 1: then it goes. And I really do think about computer

344
00:16:53,440 --> 00:16:56,360
Speaker 1: programs the way you might think about little machines, right,

345
00:16:56,480 --> 00:16:59,080
Speaker 1: because that's exactly what they are. They just execute a

346
00:16:59,120 --> 00:17:01,200
Speaker 1: set of instructions. You know. It's just like a bunch

347
00:17:01,200 --> 00:17:04,159
Speaker 1: of gears clicking into place, and they can't change the

348
00:17:04,200 --> 00:17:05,879
Speaker 1: way they do that. And they do it regardless of

349
00:17:05,880 --> 00:17:08,119
Speaker 1: whether it's the right thing, or whether it's effective or whatever.

350
00:17:08,160 --> 00:17:08,800
Speaker 1: It just goes.

351
00:17:09,119 --> 00:17:13,800
Speaker 3: Hmmm, like your electric toothbrush, you know, you switch it on,

352
00:17:14,000 --> 00:17:16,080
Speaker 3: and it's just that it has a circuit that just

353
00:17:16,320 --> 00:17:18,760
Speaker 3: has it moved the bristles back and forth.

354
00:17:18,880 --> 00:17:20,639
Speaker 1: That's right. And it doesn't know if it's brushing your

355
00:17:20,640 --> 00:17:22,679
Speaker 1: teeth or just flailing around in mid air. Right, It

356
00:17:22,720 --> 00:17:24,840
Speaker 1: has no idea, It doesn't care, it doesn't think or

357
00:17:24,840 --> 00:17:27,199
Speaker 1: feel whatever. It's just a machine, right, thank god. It

358
00:17:27,200 --> 00:17:31,320
Speaker 1: doesn't know. It would tell you to brush your teeth

359
00:17:31,400 --> 00:17:36,439
Speaker 1: off word like oh, less chocolate. Whohe, I'm tired of this?

360
00:17:37,640 --> 00:17:38,520
Speaker 3: What is this gunk?

361
00:17:38,680 --> 00:17:40,520
Speaker 1: Yeah? Exactly, And so that's what a sort of a

362
00:17:40,560 --> 00:17:43,199
Speaker 1: normal machine is. That's what like a classical computer program is. Right.

363
00:17:43,240 --> 00:17:44,680
Speaker 1: Think of it just the same way as you think

364
00:17:44,680 --> 00:17:45,720
Speaker 1: of a physical machine.

365
00:17:45,800 --> 00:17:49,719
Speaker 3: Okay, it's just doing what you the programmer told it

366
00:17:49,760 --> 00:17:50,000
Speaker 3: to do.

367
00:17:50,280 --> 00:17:54,199
Speaker 1: That's right, and it follows your instructions exactly. Now, a

368
00:17:54,240 --> 00:17:56,919
Speaker 1: computer that can learn is different, right. A computer that

369
00:17:56,960 --> 00:17:59,840
Speaker 1: has artificial intelligence is different in this really important way

370
00:18:00,160 --> 00:18:03,760
Speaker 1: because you can train it right, and you can train

371
00:18:03,840 --> 00:18:07,320
Speaker 1: it because we build these things to model the way

372
00:18:07,359 --> 00:18:11,320
Speaker 1: that we work. Right. So, for example, an AI program

373
00:18:11,359 --> 00:18:14,760
Speaker 1: is sort of like like a newborn baby can't do anything.

374
00:18:15,000 --> 00:18:15,200
Speaker 6: Right.

375
00:18:15,600 --> 00:18:18,199
Speaker 1: Say, there's a AI program, for example, that's supposed to

376
00:18:18,640 --> 00:18:21,679
Speaker 1: recognize you when you come in the door. Right, is

377
00:18:21,720 --> 00:18:24,320
Speaker 1: this Orge or is this not Jorge? Right? Because I

378
00:18:24,400 --> 00:18:26,720
Speaker 1: should only open the door for Joorge and not open

379
00:18:26,760 --> 00:18:30,080
Speaker 1: the door for not Jorge. Okay. So when you create

380
00:18:30,440 --> 00:18:32,680
Speaker 1: a new AI program, you would start out like just

381
00:18:32,760 --> 00:18:34,440
Speaker 1: a newborn baby, okay.

382
00:18:34,160 --> 00:18:36,639
Speaker 3: And like a blank slate, right, yeah, like a.

383
00:18:36,600 --> 00:18:39,200
Speaker 1: Blank slate, and it would make random decisions, right, You

384
00:18:39,880 --> 00:18:41,959
Speaker 1: show it a face and it would say yes, it's

385
00:18:42,080 --> 00:18:44,600
Speaker 1: orge and then you say no you were wrong, or

386
00:18:44,680 --> 00:18:47,359
Speaker 1: yes you were right, and then you would reward it

387
00:18:47,400 --> 00:18:49,600
Speaker 1: if it does well, if it gives the right answer,

388
00:18:49,960 --> 00:18:53,280
Speaker 1: and you would you would punish it. If it doesn't right,

389
00:18:53,320 --> 00:18:55,280
Speaker 1: you would tell it. I mean you don't actually punish

390
00:18:55,359 --> 00:18:57,000
Speaker 1: it or reward it. You just tell it, yes, you

391
00:18:57,160 --> 00:18:59,639
Speaker 1: made the right call this time, and know you made

392
00:18:59,680 --> 00:19:01,720
Speaker 1: the wrong call this time this other time.

393
00:19:01,800 --> 00:19:04,639
Speaker 3: But how is that different than the idea of calibrating something?

394
00:19:04,880 --> 00:19:06,960
Speaker 3: Do you know what I mean? Like it's calibration then

395
00:19:07,200 --> 00:19:08,320
Speaker 3: artificial intelligence.

396
00:19:08,680 --> 00:19:12,040
Speaker 1: Right, Well, the difference is calibration is like here, I

397
00:19:12,080 --> 00:19:14,000
Speaker 1: have a tool. I know how to solve the problem.

398
00:19:14,000 --> 00:19:16,040
Speaker 1: I just have to adjust it so that it does

399
00:19:16,080 --> 00:19:18,600
Speaker 1: exactly the right thing here, right, But you have a

400
00:19:18,600 --> 00:19:22,199
Speaker 1: strategy that it's executing. You know. It's like you have

401
00:19:22,240 --> 00:19:25,000
Speaker 1: a drill and you want it to drill fast or slow,

402
00:19:25,040 --> 00:19:26,800
Speaker 1: and you know you know how to solve the problem.

403
00:19:26,840 --> 00:19:29,040
Speaker 1: You just you know it has to spin and screw

404
00:19:29,119 --> 00:19:32,160
Speaker 1: the thing in or whatever. It's just adjusting a knob here.

405
00:19:32,240 --> 00:19:34,320
Speaker 1: You don't know how to solve the problem, and so

406
00:19:34,400 --> 00:19:38,080
Speaker 1: you've given it a very very very flexible strategy on

407
00:19:38,600 --> 00:19:40,920
Speaker 1: the inside you've given it. Like, imagine something has like

408
00:19:40,960 --> 00:19:44,800
Speaker 1: a thousand knobs, right, If you twist all these knobs,

409
00:19:44,840 --> 00:19:47,480
Speaker 1: you could get all sorts of crazy strategies.

410
00:19:47,960 --> 00:19:48,080
Speaker 6: Right.

411
00:19:48,119 --> 00:19:50,760
Speaker 1: So back to the example of like recognizing whoorhe or not,

412
00:19:51,880 --> 00:19:55,320
Speaker 1: when you tell it it's done, it's given the wrong answer,

413
00:19:55,640 --> 00:19:58,399
Speaker 1: then it adjusts those knobs. It says, well, let me

414
00:19:58,440 --> 00:20:01,160
Speaker 1: try to tweak my strategy for siding is this hooge,

415
00:20:01,560 --> 00:20:02,800
Speaker 1: and then we'll see how that goes.

416
00:20:03,400 --> 00:20:06,680
Speaker 3: I think that's a key difference. It's the number of knobs, right,

417
00:20:06,760 --> 00:20:09,480
Speaker 3: like a drill with the knob for velocity. I mean

418
00:20:09,480 --> 00:20:12,080
Speaker 3: that is sort of trainable, and you can set it

419
00:20:12,160 --> 00:20:15,080
Speaker 3: up to be adaptive, but it's just one knob, and

420
00:20:15,119 --> 00:20:18,960
Speaker 3: so it's not. You wouldn't say it's intelligent. It's not intelligent.

421
00:20:19,080 --> 00:20:22,000
Speaker 3: The spectrum of things it can do is very very small. Yeah, right.

422
00:20:22,040 --> 00:20:24,840
Speaker 3: But whereas like something that recognizes a face, it needs

423
00:20:24,880 --> 00:20:28,360
Speaker 3: to evaluate like a million pixels in a photo, right,

424
00:20:28,640 --> 00:20:30,879
Speaker 3: mm hmm, And so for you to tweak how it

425
00:20:31,080 --> 00:20:34,720
Speaker 3: evaluates each of those pixels, it would be really difficult.

426
00:20:34,400 --> 00:20:36,800
Speaker 1: For you to That's right. So imagine the machine here

427
00:20:36,880 --> 00:20:38,480
Speaker 1: is is a camera in the door and takes the

428
00:20:38,480 --> 00:20:40,479
Speaker 1: picture of hooreey. You got a million pixels, and then

429
00:20:40,520 --> 00:20:42,399
Speaker 1: it has to look at those pixels and decide is

430
00:20:42,440 --> 00:20:45,080
Speaker 1: this Joge or is this not Jorge? And so there's

431
00:20:45,320 --> 00:20:48,080
Speaker 1: does some calculation on that picture, right, and that calculation

432
00:20:48,240 --> 00:20:50,959
Speaker 1: has millions of knobs on it, Right, how much door

433
00:20:50,960 --> 00:20:53,440
Speaker 1: I weigh this pixel? How much do I weigh adjacent pixels?

434
00:20:53,560 --> 00:20:55,000
Speaker 1: Do I look for his nose? Do I look for

435
00:20:55,040 --> 00:20:57,399
Speaker 1: his hair? Do I look for the eyes? Right? So

436
00:20:57,440 --> 00:21:00,320
Speaker 1: it's got some very very flexible thing inside of it

437
00:21:00,320 --> 00:21:02,520
Speaker 1: that can do almost anything. And when you first start out,

438
00:21:02,640 --> 00:21:05,840
Speaker 1: it's just random. So it's making ridiculous, terrible decisions. But

439
00:21:05,920 --> 00:21:08,840
Speaker 1: the key the thing that models the learning, right, you know,

440
00:21:08,960 --> 00:21:12,040
Speaker 1: you just need artificial intelligence. You need artificial learning. The

441
00:21:12,119 --> 00:21:14,680
Speaker 1: thing that models that learning is that when it gets

442
00:21:14,720 --> 00:21:17,760
Speaker 1: the wrong answer, it knows how to adjust those knobs

443
00:21:18,000 --> 00:21:20,800
Speaker 1: so that next time it's more correct by itself.

444
00:21:20,920 --> 00:21:23,160
Speaker 3: Yeah, that's the key thing is that it learns by itself.

445
00:21:23,160 --> 00:21:26,400
Speaker 3: It doesn't need you. They're sitting like, oh, you got

446
00:21:26,400 --> 00:21:28,639
Speaker 3: this pixel wrong, you got that pixel wrong. I mean

447
00:21:28,680 --> 00:21:31,160
Speaker 3: to tweak, you know, tweak this one this way. It's

448
00:21:31,240 --> 00:21:34,479
Speaker 3: really more like an eponymous automatic learning.

449
00:21:34,720 --> 00:21:36,560
Speaker 1: That's right because you don't know how to adjust it.

450
00:21:36,560 --> 00:21:37,800
Speaker 1: If you knew how to adjust it, you would just

451
00:21:37,840 --> 00:21:40,919
Speaker 1: write that program. Right. The key is artificial intelligence is

452
00:21:40,960 --> 00:21:43,840
Speaker 1: excellent when you don't know how to solve the problem,

453
00:21:43,960 --> 00:21:46,360
Speaker 1: but you can define the problem. You can say this

454
00:21:46,400 --> 00:21:47,720
Speaker 1: is a picture of or he and this is not

455
00:21:48,480 --> 00:21:51,159
Speaker 1: learn a way to tell the difference, right. So you

456
00:21:51,160 --> 00:21:54,399
Speaker 1: give it a very flexible strategy and then you try

457
00:21:54,560 --> 00:21:56,439
Speaker 1: You let it try out, and when it gives it

458
00:21:56,480 --> 00:21:58,919
Speaker 1: the wrong answer, you would let it adjust itself so

459
00:21:58,960 --> 00:22:01,200
Speaker 1: that it gets closer and clar to giving the right answer,

460
00:22:01,520 --> 00:22:04,720
Speaker 1: and eventually these things will find the right setting for

461
00:22:04,760 --> 00:22:07,760
Speaker 1: those millions of knobs, so that it's doing the right thing.

462
00:22:07,760 --> 00:22:09,800
Speaker 1: It's saying, oh, look, this picture is a picture of

463
00:22:09,840 --> 00:22:12,040
Speaker 1: wohe and it gets the right answer ninety nine percent

464
00:22:12,040 --> 00:22:13,600
Speaker 1: of the time. And when you give it a picture

465
00:22:13,640 --> 00:22:15,520
Speaker 1: that's not a picture of horehey and you give it Daniel,

466
00:22:15,720 --> 00:22:18,720
Speaker 1: it says, no, sorry, you're not getting in the house right.

467
00:22:19,119 --> 00:22:21,360
Speaker 3: And I think a key thing is also that you,

468
00:22:21,400 --> 00:22:25,440
Speaker 3: as a programmer, could not have predicted what all those

469
00:22:25,480 --> 00:22:27,399
Speaker 3: knobs are going to be at the end, right, Like

470
00:22:27,440 --> 00:22:30,000
Speaker 3: it's such a big problem. There's a million knobs. There's

471
00:22:30,040 --> 00:22:33,040
Speaker 3: no way that you can predict what those knobs are

472
00:22:33,080 --> 00:22:35,119
Speaker 3: going to be set to when it learns my face.

473
00:22:35,359 --> 00:22:38,000
Speaker 1: That's right. It's perfect for really hard problems where we

474
00:22:38,040 --> 00:22:41,000
Speaker 1: don't know how to solve it. Right. We know how

475
00:22:41,000 --> 00:22:42,560
Speaker 1: to describe the problem, but we don't know how to

476
00:22:42,560 --> 00:22:44,679
Speaker 1: solve it. You're right. So if I already knew how

477
00:22:44,680 --> 00:22:46,280
Speaker 1: to solve it, I could write a computer program that

478
00:22:47,119 --> 00:22:49,840
Speaker 1: and tell it just like use this pixel, use that pixel,

479
00:22:49,920 --> 00:22:51,640
Speaker 1: use this pixel. But I don't know how to solve

480
00:22:51,640 --> 00:22:54,439
Speaker 1: that problem. It's really hard, right, But I can train

481
00:22:54,640 --> 00:22:57,920
Speaker 1: a computer to figure it out, just the same way

482
00:22:58,160 --> 00:23:01,200
Speaker 1: I can train a dog. Right. A dog can learn

483
00:23:01,240 --> 00:23:04,719
Speaker 1: my face. Right, a dog recognizes its owner, and you know,

484
00:23:04,760 --> 00:23:08,200
Speaker 1: happily licks its face when it comes home, and recognizes

485
00:23:08,240 --> 00:23:10,760
Speaker 1: that when somebody's not its owner, and barks like crazy

486
00:23:10,760 --> 00:23:12,840
Speaker 1: and choose its face off when it's not its owner.

487
00:23:12,920 --> 00:23:17,320
Speaker 3: Right, Remind me not to visit your house, Daniel. Seems

488
00:23:17,320 --> 00:23:23,959
Speaker 3: a little dangerous. So then a big thing is programming

489
00:23:24,080 --> 00:23:27,160
Speaker 3: a structure in the in the software that is kind

490
00:23:27,200 --> 00:23:30,440
Speaker 3: of open ended and malleable, do you know what I mean?

491
00:23:30,480 --> 00:23:33,560
Speaker 3: Like it that's right, something that is kind of unpredictable

492
00:23:33,560 --> 00:23:35,320
Speaker 3: in a way that can learn.

493
00:23:35,520 --> 00:23:37,399
Speaker 1: That's right. And that's a key thing is that some

494
00:23:37,440 --> 00:23:39,200
Speaker 1: people might be thinking, well, hold on, you said the

495
00:23:39,200 --> 00:23:41,480
Speaker 1: computers can just do what they tell you, So how

496
00:23:41,480 --> 00:23:42,480
Speaker 1: can a computer learn?

497
00:23:42,560 --> 00:23:42,639
Speaker 8: Right?

498
00:23:42,720 --> 00:23:45,000
Speaker 1: How's that possible? And the key is that it's an

499
00:23:45,000 --> 00:23:48,119
Speaker 1: emergent property. Right. Like the way that you write a

500
00:23:48,119 --> 00:23:50,800
Speaker 1: computer program that can learn is you build all these

501
00:23:50,840 --> 00:23:54,760
Speaker 1: little calculating bits with knobs on them, right, and each

502
00:23:54,800 --> 00:23:57,320
Speaker 1: bit just does what it's told. It takes some data,

503
00:23:57,600 --> 00:23:59,640
Speaker 1: it makes a decision based on the value of the knob,

504
00:23:59,680 --> 00:24:02,840
Speaker 1: and it's sends out some data and together all these

505
00:24:02,880 --> 00:24:05,520
Speaker 1: things make a decision. Right. Each individual piece has no

506
00:24:05,560 --> 00:24:07,800
Speaker 1: idea what it's doing. It's not smart or intelligent or

507
00:24:07,800 --> 00:24:10,520
Speaker 1: making its own decisions. That doesn't have free will, right,

508
00:24:10,800 --> 00:24:14,240
Speaker 1: But together they're doing something. And as you said earlier,

509
00:24:14,400 --> 00:24:16,920
Speaker 1: they can change the way they behave They can adjust

510
00:24:16,920 --> 00:24:20,080
Speaker 1: these knobs themselves to improve their performance. That's where the

511
00:24:20,160 --> 00:24:22,920
Speaker 1: learning comes from. It's from that training. It gets external

512
00:24:22,960 --> 00:24:25,680
Speaker 1: input and changes its behavior based on that exp.

513
00:24:25,720 --> 00:24:28,000
Speaker 3: I see you're saying that the way to program these

514
00:24:28,040 --> 00:24:33,160
Speaker 3: ais is by connecting a bunch of little simple things

515
00:24:33,160 --> 00:24:35,280
Speaker 3: together to get something complex.

516
00:24:35,440 --> 00:24:37,679
Speaker 1: Yes, And here it's important to remember that we are

517
00:24:37,800 --> 00:24:40,320
Speaker 1: using neural networks as sort of a stand in to

518
00:24:40,359 --> 00:24:43,320
Speaker 1: represent a big broad set of strategies that are part

519
00:24:43,320 --> 00:24:46,199
Speaker 1: of machine learning, right, And you don't know how to

520
00:24:46,240 --> 00:24:47,959
Speaker 1: set them, how to put them together to get the

521
00:24:48,000 --> 00:24:50,840
Speaker 1: right complex behavior. You just put them together and then

522
00:24:50,880 --> 00:24:53,520
Speaker 1: you train it. Right. You say, well, I have something

523
00:24:53,600 --> 00:24:55,760
Speaker 1: that's dumb, like a newborn baby, and I teach it

524
00:24:55,960 --> 00:24:57,120
Speaker 1: how to do the thing that I want.

525
00:24:57,200 --> 00:25:00,919
Speaker 3: But this really all sort of came about from brain research, right, Like,

526
00:25:00,960 --> 00:25:02,879
Speaker 3: people were studying the brain and they figured out that

527
00:25:02,960 --> 00:25:05,440
Speaker 3: our brains are made up of all these little simple

528
00:25:05,520 --> 00:25:09,240
Speaker 3: units neurons, that's right. And each neuron is pretty simple, right, Like,

529
00:25:09,280 --> 00:25:11,639
Speaker 3: it just takes a couple of inputs and then it

530
00:25:11,800 --> 00:25:13,640
Speaker 3: just outputs one signal.

531
00:25:13,840 --> 00:25:16,359
Speaker 1: That's the really fascinating deep part about it, right, is

532
00:25:16,359 --> 00:25:19,480
Speaker 1: that the structures we use in computers are modeled after

533
00:25:19,600 --> 00:25:23,360
Speaker 1: what's actually happening in real brains. And you say, inside

534
00:25:23,359 --> 00:25:25,520
Speaker 1: your brain are a bunch of neurons, right, And these

535
00:25:25,560 --> 00:25:28,920
Speaker 1: neurons taken some input and then if the input is

536
00:25:29,000 --> 00:25:31,280
Speaker 1: right or above some a certain amount, then they send

537
00:25:31,359 --> 00:25:33,880
Speaker 1: us some output, which is the input to the next neuron, right,

538
00:25:33,920 --> 00:25:35,879
Speaker 1: and your brain is basically just a big web of

539
00:25:35,920 --> 00:25:36,440
Speaker 1: these things.

540
00:25:36,520 --> 00:25:39,840
Speaker 3: Yeah, yeah, that's right. That's the key is that these neurons,

541
00:25:39,920 --> 00:25:42,520
Speaker 3: they're simple, but they're all sort of connected to each other.

542
00:25:43,000 --> 00:25:46,200
Speaker 3: So it's a huge complex web going on inside your head.

543
00:25:46,640 --> 00:25:49,120
Speaker 3: And when you're learning, what you're doing is you're kind

544
00:25:49,119 --> 00:25:52,080
Speaker 3: of like shaping that web. You're saying, some connections. These

545
00:25:52,080 --> 00:25:57,479
Speaker 3: connections are important for recognizing core. These connections are important

546
00:25:57,520 --> 00:25:59,800
Speaker 3: when you want to when it's not hore kind of thing.

547
00:26:00,080 --> 00:26:00,359
Speaker 6: That's right.

548
00:26:00,400 --> 00:26:03,119
Speaker 1: Your neurons can change. They have like basically knobs on them.

549
00:26:03,119 --> 00:26:06,160
Speaker 1: I mean not physical literal knobs, but they have they

550
00:26:06,160 --> 00:26:09,080
Speaker 1: can adjust. And so if you feel pain, you know,

551
00:26:09,200 --> 00:26:12,280
Speaker 1: or you have an experience, then that changes the way

552
00:26:12,320 --> 00:26:14,960
Speaker 1: your neurons work and it changes a little bit who

553
00:26:14,960 --> 00:26:16,680
Speaker 1: you are and how you react to things. And that's

554
00:26:16,720 --> 00:26:20,119
Speaker 1: why you know, newborn babies when they're born, they're not

555
00:26:20,359 --> 00:26:24,480
Speaker 1: very responsive to stimuli because they're just still figuring it out.

556
00:26:24,480 --> 00:26:26,639
Speaker 1: You know, a newborn baby doesn't even know like, this

557
00:26:26,800 --> 00:26:28,719
Speaker 1: is my arm, and I know how to control it.

558
00:26:28,720 --> 00:26:32,159
Speaker 1: Has to learn all of these things by being trained

559
00:26:32,200 --> 00:26:33,639
Speaker 1: by having experiences.

560
00:26:33,880 --> 00:26:37,040
Speaker 3: You know, it has the neurons, and the neurons are

561
00:26:37,119 --> 00:26:39,640
Speaker 3: connected to each other, but it hasn't figure out how

562
00:26:39,640 --> 00:26:40,680
Speaker 3: to use those connections.

563
00:26:40,920 --> 00:26:42,920
Speaker 1: That's right. It has to be trained to be useful

564
00:26:43,280 --> 00:26:45,000
Speaker 1: and to interact with the world in any sort of

565
00:26:45,040 --> 00:26:48,960
Speaker 1: meaningful way, right, And so that's it's exactly the same sense.

566
00:26:48,960 --> 00:26:51,760
Speaker 1: And it's fascinating that if you build a mathematical system

567
00:26:51,800 --> 00:26:54,639
Speaker 1: that's what a computer program is, basically a mathematical model

568
00:26:54,920 --> 00:26:57,520
Speaker 1: of the processes that are happening in your brain. It

569
00:26:57,560 --> 00:26:59,600
Speaker 1: performs in a very similar way, and it does this

570
00:26:59,680 --> 00:27:02,880
Speaker 1: amazing thing, which is it adjusts itself to improve its

571
00:27:02,920 --> 00:27:05,600
Speaker 1: performance on the task you've given it. Right, So it

572
00:27:05,680 --> 00:27:08,200
Speaker 1: really is like a model of learning. And when people

573
00:27:08,240 --> 00:27:10,800
Speaker 1: saw this, they said, wow, I mean you look inside

574
00:27:10,800 --> 00:27:13,439
Speaker 1: the brain. You're wondering, like, how does thinking work? Where's

575
00:27:13,480 --> 00:27:13,879
Speaker 1: the soul?

576
00:27:14,080 --> 00:27:14,199
Speaker 4: Right?

577
00:27:14,240 --> 00:27:16,159
Speaker 1: Where am I? You look inside the brain? All you

578
00:27:16,160 --> 00:27:18,119
Speaker 1: see all these weird neurons connected to each other, and

579
00:27:18,160 --> 00:27:21,600
Speaker 1: you think, how could that possibly describe me? But when

580
00:27:21,600 --> 00:27:23,199
Speaker 1: you build a model of it in a computer and

581
00:27:23,240 --> 00:27:25,280
Speaker 1: it can do the things that you can do, which

582
00:27:25,320 --> 00:27:27,560
Speaker 1: is you learn and develop and react.

583
00:27:27,200 --> 00:27:29,119
Speaker 3: And be trained, yeah, and make back jokes.

584
00:27:30,640 --> 00:27:34,399
Speaker 1: Not yet. We have not yet solved the bad joke problem.

585
00:27:34,480 --> 00:27:34,560
Speaker 6: Right.

586
00:27:34,680 --> 00:27:36,960
Speaker 1: Humans are still world champions in terms of bad jobs.

587
00:27:38,280 --> 00:27:40,320
Speaker 3: We can still beat them at something that's right.

588
00:27:40,359 --> 00:27:42,320
Speaker 1: And you know, and this is very useful because you

589
00:27:42,400 --> 00:27:45,359
Speaker 1: want the systems around you to learn and to react,

590
00:27:45,400 --> 00:27:49,280
Speaker 1: you know. And like if your phone, for example, it knows, hey,

591
00:27:49,400 --> 00:27:52,080
Speaker 1: every time you open your phone, you start with Twitter, right,

592
00:27:52,119 --> 00:27:54,720
Speaker 1: and so Twitter goes up on there on the most

593
00:27:54,920 --> 00:27:57,920
Speaker 1: used app list, right. And that's not a very complex

594
00:27:58,000 --> 00:28:01,640
Speaker 1: artificial intelligence, but it is. And these sort of things

595
00:28:01,680 --> 00:28:02,359
Speaker 1: are very helpful.

596
00:28:02,800 --> 00:28:03,800
Speaker 3: Let's take a quick break.

597
00:28:08,240 --> 00:28:10,040
Speaker 1: When you pop a piece of cheese into your mouth,

598
00:28:10,160 --> 00:28:13,280
Speaker 1: or enjoy a rich spoonful of Greek yogurt, you're probably

599
00:28:13,320 --> 00:28:17,359
Speaker 1: not thinking about the environmental impact of each and every bite.

600
00:28:17,400 --> 00:28:20,040
Speaker 1: But the people in the dairy industry are US. Dairy

601
00:28:20,080 --> 00:28:24,359
Speaker 1: has set themselves some ambitious sustainability goals, including being greenhouse

602
00:28:24,400 --> 00:28:26,960
Speaker 1: gas neutral by twenty to fifty. That's why they're working

603
00:28:27,000 --> 00:28:29,320
Speaker 1: hard every day to find new ways to reduce waste,

604
00:28:29,400 --> 00:28:33,640
Speaker 1: conserve natural resources, and drive down greenhouse gas emissions. Take water,

605
00:28:33,680 --> 00:28:36,760
Speaker 1: for example, most dairy farms reuse water up to four

606
00:28:36,800 --> 00:28:40,280
Speaker 1: times the same water cools the milk, cleans equipment, washes

607
00:28:40,320 --> 00:28:43,120
Speaker 1: the barn, and irrigates the crops. How is US Dairy

608
00:28:43,160 --> 00:28:46,880
Speaker 1: tackling greenhouse gases? Many farms use anaerobic digestors that turn

609
00:28:46,960 --> 00:28:50,840
Speaker 1: the methane from maneuver into renewable energy that can power farms, towns,

610
00:28:50,880 --> 00:28:52,960
Speaker 1: and electric cars. So the next time you grab a

611
00:28:53,000 --> 00:28:55,000
Speaker 1: slice of pizza or lick an ice cream cone, know

612
00:28:55,080 --> 00:28:57,760
Speaker 1: that dairy farmers and processors around the country are using

613
00:28:57,800 --> 00:29:01,280
Speaker 1: the latest practices and innovations to provide the nutrient intents

614
00:29:01,400 --> 00:29:04,120
Speaker 1: dairy products we love with less of an impact. Visit

615
00:29:04,240 --> 00:29:07,040
Speaker 1: usdairy dot com slash sustainability to learn more.

616
00:29:07,440 --> 00:29:10,480
Speaker 8: This episode is brought to you by Navy Federal Credit Union.

617
00:29:10,840 --> 00:29:14,160
Speaker 8: Buying a home in today's market can be overwhelming. Luckily,

618
00:29:14,360 --> 00:29:17,360
Speaker 8: Navy Federal Credit Union's new Home Buying Center has everything

619
00:29:17,400 --> 00:29:19,760
Speaker 8: you need to keep you sane while you shop. With

620
00:29:19,920 --> 00:29:23,760
Speaker 8: services like Verified Preapproval, which show sellers you're serious buyer,

621
00:29:24,000 --> 00:29:27,000
Speaker 8: you have a competitive advantage when making an offer, and

622
00:29:27,160 --> 00:29:29,520
Speaker 8: lock and Shop lets you lock in your interest rate

623
00:29:29,560 --> 00:29:31,360
Speaker 8: for up to sixty days while you hunt for a

624
00:29:31,440 --> 00:29:34,680
Speaker 8: home from start to finish. Navy Federal's new Home Buying

625
00:29:34,720 --> 00:29:37,600
Speaker 8: Center has everything you'll need to buy a home. Navy

626
00:29:37,600 --> 00:29:40,920
Speaker 8: Federal Credit Union Our members are the mission. Learn more

627
00:29:40,960 --> 00:29:44,000
Speaker 8: at Navy Federal dot org. Navy Federal is ensured by

628
00:29:44,120 --> 00:29:47,600
Speaker 8: NCUA Equal Housing Lender Membership Required terms and conditions apply.

629
00:29:47,720 --> 00:29:50,320
Speaker 8: Loan subject to approval. Call one eight eight eight four

630
00:29:50,320 --> 00:29:52,680
Speaker 8: two sixty three two eight for details about credit costs.

631
00:29:54,080 --> 00:29:57,600
Speaker 9: There are children, friends and families walking, riding on passing

632
00:29:57,640 --> 00:29:59,960
Speaker 9: the roads every day. Remember their real people with love,

633
00:30:00,000 --> 00:30:02,120
Speaker 9: the ones who need them to get home safely. Protect

634
00:30:02,160 --> 00:30:04,360
Speaker 9: our cyclists and pedestrians because they're people too.

635
00:30:04,640 --> 00:30:05,200
Speaker 1: Go safely.

636
00:30:05,240 --> 00:30:08,120
Speaker 9: California from the California Office of Traffic Safety and Caltrans.

637
00:30:16,480 --> 00:30:18,520
Speaker 3: So that's kind of what makes AI is that a

638
00:30:18,960 --> 00:30:22,160
Speaker 3: it can tackle complex problems that we wouldn't even know

639
00:30:22,160 --> 00:30:25,280
Speaker 3: how to program something to do. And b that it

640
00:30:25,360 --> 00:30:28,560
Speaker 3: changes and adapts and kind of it can get better,

641
00:30:28,840 --> 00:30:31,000
Speaker 3: not just better, but also kind of adapt to the

642
00:30:31,040 --> 00:30:31,800
Speaker 3: person using it.

643
00:30:32,120 --> 00:30:35,440
Speaker 1: That's exactly right, exactly right. And so for example, sometimes

644
00:30:35,440 --> 00:30:38,920
Speaker 1: you know Netflix uses AI. It says what program will

645
00:30:39,000 --> 00:30:42,840
Speaker 1: you want to watch next? Well, you know that's an AI.

646
00:30:42,920 --> 00:30:45,760
Speaker 1: It's been trained, they feeded a bunch of examples. They say,

647
00:30:46,120 --> 00:30:49,080
Speaker 1: Bob watched these five shows, and then he watched this

648
00:30:49,280 --> 00:30:51,880
Speaker 1: sixth show. But they gave the AI just the first

649
00:30:51,960 --> 00:30:54,480
Speaker 1: five and they ask it predict what show he will

650
00:30:54,520 --> 00:30:56,920
Speaker 1: watch next, and then they see if it doesn't a

651
00:30:56,960 --> 00:30:58,440
Speaker 1: good job, and if it does a good job. You know,

652
00:30:58,480 --> 00:31:00,200
Speaker 1: they were warned. If it does a bad job, it

653
00:31:00,240 --> 00:31:03,520
Speaker 1: seems its knobs to do better. Then when you're sitting

654
00:31:03,560 --> 00:31:06,040
Speaker 1: there watching five hours of Netflix, it can do a

655
00:31:06,040 --> 00:31:08,280
Speaker 1: pretty good job of predicting what you're going to watch

656
00:31:08,360 --> 00:31:11,120
Speaker 1: next because it's been trained on a lot of data.

657
00:31:11,240 --> 00:31:13,720
Speaker 1: This is why people always talking about big data. Big data.

658
00:31:14,040 --> 00:31:16,240
Speaker 1: These companies are gathering data about you so they can

659
00:31:16,320 --> 00:31:19,800
Speaker 1: train their ais to learn your behavior and predict it.

660
00:31:20,920 --> 00:31:23,480
Speaker 3: Except the problem is me and my spouse we share

661
00:31:23,560 --> 00:31:26,800
Speaker 3: the same account and the same log in, so.

662
00:31:27,240 --> 00:31:30,640
Speaker 1: That's right, So it's learning some weird combination of your

663
00:31:31,080 --> 00:31:31,880
Speaker 1: in your wife's brain.

664
00:31:31,920 --> 00:31:34,320
Speaker 3: I have a very confused Netflix account.

665
00:31:34,960 --> 00:31:36,840
Speaker 1: Or maybe it understands your marriage better.

666
00:31:36,680 --> 00:31:40,960
Speaker 3: Than maybe he's trying to tell us something. It's like, you, guys,

667
00:31:40,960 --> 00:31:42,320
Speaker 3: should you guys watching?

668
00:31:42,400 --> 00:31:44,560
Speaker 1: Every time his wife is out of town, he watches

669
00:31:44,680 --> 00:31:47,320
Speaker 1: these shows. When she's in town, he has to watch

670
00:31:47,400 --> 00:31:48,560
Speaker 1: these other shows.

671
00:31:49,360 --> 00:31:57,920
Speaker 3: Oh oh, all right, Dine break it down for us.

672
00:31:58,160 --> 00:32:01,960
Speaker 3: How long before the AI is take over the world?

673
00:32:03,840 --> 00:32:06,520
Speaker 1: Not very long actually, But you're asking a different question earlier,

674
00:32:06,520 --> 00:32:09,480
Speaker 1: which is is AI dangerous? And I think that has

675
00:32:09,480 --> 00:32:10,520
Speaker 1: two different questions there.

676
00:32:10,440 --> 00:32:12,840
Speaker 3: Right, I mean people are concerned. Some people are concerned.

677
00:32:13,600 --> 00:32:15,320
Speaker 1: Yeah, I think people are concerned, and they're a good

678
00:32:15,360 --> 00:32:17,520
Speaker 1: reason for it to be concerned. You know, One question

679
00:32:17,680 --> 00:32:21,440
Speaker 1: is will AI develop its own autonomy and uh and

680
00:32:21,600 --> 00:32:23,760
Speaker 1: you know take over. That's a different question from are

681
00:32:23,760 --> 00:32:25,880
Speaker 1: they dangerous, because you know they could take over and

682
00:32:25,920 --> 00:32:28,040
Speaker 1: then take better care of the planet than we have,

683
00:32:28,200 --> 00:32:31,680
Speaker 1: in which case you know they're not dangerous, they're benevolent dictators. Oh.

684
00:32:31,720 --> 00:32:34,600
Speaker 1: I think the real question is will they take over?

685
00:32:34,600 --> 00:32:37,480
Speaker 1: Will they become autonomous we lose control of them somehow?

686
00:32:37,760 --> 00:32:39,440
Speaker 1: And could they become smarter than us?

687
00:32:39,560 --> 00:32:39,720
Speaker 2: Oh?

688
00:32:39,800 --> 00:32:43,160
Speaker 3: I see, it's two issues. One could it? Could it

689
00:32:43,240 --> 00:32:46,080
Speaker 3: develop a consciousness on its own? And be is that

690
00:32:46,160 --> 00:32:48,280
Speaker 3: consciousness good or bad for us? H?

691
00:32:48,720 --> 00:32:50,880
Speaker 1: And it's an important question because as soon as you

692
00:32:50,920 --> 00:32:55,400
Speaker 1: identify learning with consciousness, right, then you wonder about that

693
00:32:55,480 --> 00:32:58,080
Speaker 1: and and this connection between the structure of AI and

694
00:32:58,120 --> 00:33:00,640
Speaker 1: the structure our brain begs that quest question. You know,

695
00:33:00,920 --> 00:33:04,080
Speaker 1: if you created, for example, an artificial hooge in the computer,

696
00:33:04,120 --> 00:33:07,520
Speaker 1: if you built a set of neurons that mimic your brain,

697
00:33:07,640 --> 00:33:11,240
Speaker 1: you know, would that simulation be alive? Would it be aware?

698
00:33:11,280 --> 00:33:14,000
Speaker 1: Would it think and would have a first person experience?

699
00:33:14,080 --> 00:33:14,280
Speaker 3: You know?

700
00:33:14,480 --> 00:33:14,600
Speaker 4: Right?

701
00:33:15,000 --> 00:33:17,920
Speaker 3: Or would it.

702
00:33:18,360 --> 00:33:20,920
Speaker 1: That's a deep philosophical question. We'll never answer it, right,

703
00:33:20,960 --> 00:33:23,400
Speaker 1: and it's not really the important question. The important question

704
00:33:23,480 --> 00:33:26,600
Speaker 1: is would we lose control of AI? Well? AI? Because

705
00:33:26,640 --> 00:33:28,960
Speaker 1: AI is something that can change and that can evolve,

706
00:33:29,280 --> 00:33:32,040
Speaker 1: It can handle complex tasks. The question is can we

707
00:33:32,120 --> 00:33:33,760
Speaker 1: lose control of it? And I think the answer to

708
00:33:33,800 --> 00:33:35,640
Speaker 1: that one is definitely yes.

709
00:33:35,760 --> 00:33:39,920
Speaker 3: We can lose control, meaning like we'll give it control

710
00:33:39,960 --> 00:33:41,400
Speaker 3: and then not be able to take it back.

711
00:33:41,800 --> 00:33:45,560
Speaker 1: Yes, exactly, because the way AI is moving is that

712
00:33:45,600 --> 00:33:47,840
Speaker 1: it can handle more and more complex tasks so that

713
00:33:47,920 --> 00:33:50,440
Speaker 1: you don't have to be super specific about what it's doing,

714
00:33:50,560 --> 00:33:54,000
Speaker 1: you know, like we have amazing natural language processing. Now

715
00:33:54,040 --> 00:33:56,160
Speaker 1: you can say sort of vague things to your phone,

716
00:33:56,560 --> 00:33:59,840
Speaker 1: like hey, set me up an appointment for tomorrow afternoon, right,

717
00:34:00,560 --> 00:34:03,600
Speaker 1: and it'll understand because it understands what your intent was. Right,

718
00:34:03,720 --> 00:34:06,160
Speaker 1: has to judge your intent and then execute it. It

719
00:34:06,240 --> 00:34:08,040
Speaker 1: used to be you have to go into your computer

720
00:34:08,080 --> 00:34:09,799
Speaker 1: and you have to press the keys and to create

721
00:34:09,800 --> 00:34:11,560
Speaker 1: that in your calendar. Now you can sort of talk

722
00:34:11,600 --> 00:34:14,520
Speaker 1: to your phone and it'll interpret what you want and

723
00:34:14,560 --> 00:34:18,000
Speaker 1: it'll do that, and that's that's awesome. That's wonderful for

724
00:34:18,160 --> 00:34:21,319
Speaker 1: human computer interactions. That we can use our language to

725
00:34:21,320 --> 00:34:23,600
Speaker 1: talk to them. We don't have to write computer code.

726
00:34:23,640 --> 00:34:26,360
Speaker 1: That's a huge step forward, right, that people can construct

727
00:34:26,480 --> 00:34:30,040
Speaker 1: machines using English rather than Python or C plus plus. Right,

728
00:34:30,320 --> 00:34:31,320
Speaker 1: it's a big step forward.

729
00:34:31,320 --> 00:34:33,399
Speaker 3: I think that's kind of what people find scary about

730
00:34:33,440 --> 00:34:36,799
Speaker 3: AI is that you can't really predict what it's going

731
00:34:36,880 --> 00:34:38,960
Speaker 3: to do. I mean, it's sort of comedy goal when

732
00:34:38,960 --> 00:34:40,600
Speaker 3: your kids are trying to talk to Alexa and ask

733
00:34:40,640 --> 00:34:43,960
Speaker 3: it funny questions. But that's kind of what's fascinating about it, right,

734
00:34:44,000 --> 00:34:47,759
Speaker 3: Like you ask it questions, you give a task, and

735
00:34:47,800 --> 00:34:50,080
Speaker 3: you really sort of don't know what it's going to do.

736
00:34:50,400 --> 00:34:53,520
Speaker 1: That's exactly right, because it's making higher and higher level decisions,

737
00:34:53,560 --> 00:34:56,000
Speaker 1: which makes it much more useful and much more intelligent.

738
00:34:56,239 --> 00:34:58,520
Speaker 1: The same way when your kid grows up, right, when

739
00:34:58,560 --> 00:35:00,400
Speaker 1: it's when your kid is four, you have to be

740
00:35:00,520 --> 00:35:02,680
Speaker 1: very specific. You have to say things out loud which

741
00:35:02,680 --> 00:35:05,840
Speaker 1: you're ridiculous, right, like don't put that finger in your nose.

742
00:35:05,960 --> 00:35:08,200
Speaker 1: You know, we're like, oh, that's been on the floor,

743
00:35:08,239 --> 00:35:09,880
Speaker 1: don't eat it. Right, you have to be really specific.

744
00:35:09,960 --> 00:35:12,600
Speaker 1: When they're ten, you can say more general things and

745
00:35:12,640 --> 00:35:13,800
Speaker 1: they'll understand.

746
00:35:13,440 --> 00:35:16,240
Speaker 3: Right, yeah, learned they're intelligent, Like don't put both fingers

747
00:35:16,239 --> 00:35:16,719
Speaker 3: in your nose.

748
00:35:18,920 --> 00:35:20,600
Speaker 1: Only put one finger in your nose out of time,

749
00:35:20,760 --> 00:35:23,239
Speaker 1: so you don't put your finger in your sisters and right,

750
00:35:24,200 --> 00:35:27,880
Speaker 1: the same way as machines or artificial intelligence gets more intelligent,

751
00:35:28,160 --> 00:35:30,759
Speaker 1: you can give it vaguan instructions and then it makes

752
00:35:30,800 --> 00:35:32,400
Speaker 1: decisions based on its.

753
00:35:32,239 --> 00:35:35,279
Speaker 3: Training, right, and you don't really know. It's just like

754
00:35:35,360 --> 00:35:38,640
Speaker 3: you can't really read what is in every neuron in

755
00:35:38,680 --> 00:35:41,399
Speaker 3: another person. In an AI, you sort of you don't

756
00:35:41,440 --> 00:35:42,520
Speaker 3: know what's going to happen, what's going.

757
00:35:42,520 --> 00:35:44,719
Speaker 1: To come out exactly. So they're going to start making

758
00:35:44,719 --> 00:35:47,480
Speaker 1: decisions based on, you know, still what we tell them

759
00:35:47,520 --> 00:35:49,640
Speaker 1: to do. But you know what if you told your AI,

760
00:35:49,680 --> 00:35:52,239
Speaker 1: you're like, hey, keep my kids safe, right, I mean,

761
00:35:52,280 --> 00:35:54,319
Speaker 1: imagine some future where you have an AI robot it's

762
00:35:54,320 --> 00:35:56,320
Speaker 1: really smart, and you say, hey, keep my kids safe,

763
00:35:56,360 --> 00:35:58,239
Speaker 1: and you come home and it's like lock them in

764
00:35:58,280 --> 00:36:01,400
Speaker 1: the basement, right, like, well, okay, they're safe. But it's

765
00:36:01,440 --> 00:36:04,040
Speaker 1: sort of a monkey pause situation, right, like you got

766
00:36:04,080 --> 00:36:06,520
Speaker 1: exactly what you asked for, but you didn't really elaborate

767
00:36:06,560 --> 00:36:08,160
Speaker 1: them right away and made different decisions.

768
00:36:08,239 --> 00:36:10,239
Speaker 3: Right, So we sort of skipped the question of whether

769
00:36:10,280 --> 00:36:13,440
Speaker 3: AIS can be, you know, a chief consciousness and become

770
00:36:13,640 --> 00:36:17,560
Speaker 3: its own kind of soul, have a soul. It doesn't

771
00:36:17,560 --> 00:36:19,160
Speaker 3: seem make you think that's a relevant question.

772
00:36:19,560 --> 00:36:21,920
Speaker 1: I think it's important because when AI gets to be

773
00:36:21,960 --> 00:36:24,600
Speaker 1: super intelligent, it's going to seem like it has a soul.

774
00:36:24,760 --> 00:36:26,719
Speaker 1: They're going to seem like people, and people are gonna

775
00:36:26,719 --> 00:36:29,880
Speaker 1: wonder like, dude, they have rights? Can you kill an AI?

776
00:36:30,320 --> 00:36:33,319
Speaker 1: What can you just delete it? You know, that's going

777
00:36:33,400 --> 00:36:35,440
Speaker 1: to be a really interesting question. But that's again, that's

778
00:36:35,440 --> 00:36:37,399
Speaker 1: a whole question of philosophy that we could We could

779
00:36:37,440 --> 00:36:39,520
Speaker 1: easily spend an hour on. Thing is a much more

780
00:36:39,520 --> 00:36:41,799
Speaker 1: practical question, which is will we lose control of them?

781
00:36:41,800 --> 00:36:44,160
Speaker 1: Whether or not they have first person experiences so they

782
00:36:44,200 --> 00:36:46,759
Speaker 1: just seem to it's important to think about whether we're

783
00:36:46,760 --> 00:36:49,360
Speaker 1: going to lose control. And there's two reasons why I

784
00:36:49,400 --> 00:36:52,280
Speaker 1: think that we will one is computers are getting faster,

785
00:36:52,680 --> 00:36:55,640
Speaker 1: really really quickly. Right, Every year computers get faster and

786
00:36:55,680 --> 00:36:58,400
Speaker 1: faster and smarter and smarter, and the scale is is

787
00:36:58,960 --> 00:37:01,640
Speaker 1: is growing, right, So this thing is happening very quickly.

788
00:37:01,680 --> 00:37:04,480
Speaker 1: But we're not right, We're not getting smarter, right, A

789
00:37:04,640 --> 00:37:06,600
Speaker 1: human brain is not changing and evolving at a very

790
00:37:06,680 --> 00:37:09,880
Speaker 1: rapid rate computers are. So they're catching up, and the

791
00:37:09,920 --> 00:37:12,600
Speaker 1: slope is step right. You just get bigger and bigger

792
00:37:12,600 --> 00:37:14,880
Speaker 1: computers and team them together and paralyze them and you

793
00:37:14,920 --> 00:37:17,920
Speaker 1: can just keep going right. So eventually they'll definitely have

794
00:37:18,320 --> 00:37:20,959
Speaker 1: enormous computing power with capabilities to do things we can't

795
00:37:20,960 --> 00:37:25,280
Speaker 1: even imagine. And also being faster doesn't necessarily mean being smarter.

796
00:37:25,360 --> 00:37:27,880
Speaker 1: You also need like more data to train on it.

797
00:37:27,920 --> 00:37:30,959
Speaker 1: And also being twice as fast doesn't mean being twice

798
00:37:31,000 --> 00:37:32,400
Speaker 1: as smart. It's not linear.

799
00:37:32,600 --> 00:37:36,839
Speaker 3: So you think that they will get more capable than us.

800
00:37:37,680 --> 00:37:40,239
Speaker 3: But do you think we will ever see control of

801
00:37:40,280 --> 00:37:43,560
Speaker 3: really important things to ais like, oh, hey, here's a

802
00:37:43,640 --> 00:37:48,360
Speaker 3: nuclear button, only fire it if it's necessary exactly?

803
00:37:48,680 --> 00:37:55,239
Speaker 1: Let's talk about weapons. Weapons is going to be what

804
00:37:55,320 --> 00:37:59,200
Speaker 1: ends it? Because you know, for example, we already have drones, right,

805
00:37:59,320 --> 00:38:01,799
Speaker 1: and we have drones with missiles on them. And these

806
00:38:01,880 --> 00:38:04,239
Speaker 1: drones can kill people. They can decide they can, you can.

807
00:38:04,400 --> 00:38:06,800
Speaker 1: Some pilot somewhere is flying it. He's making a decision

808
00:38:07,120 --> 00:38:08,919
Speaker 1: and he's going to shoot dismissile to kill a person.

809
00:38:09,000 --> 00:38:09,719
Speaker 3: Right.

810
00:38:09,800 --> 00:38:12,440
Speaker 1: But you know the enemy has drones and pretty soon

811
00:38:12,480 --> 00:38:15,279
Speaker 1: it's going to be drone on drone warfare, right, And

812
00:38:15,400 --> 00:38:17,680
Speaker 1: the drones are going to shoot each other. And at

813
00:38:17,680 --> 00:38:20,520
Speaker 1: some point somebody's going to put an AI in their drone.

814
00:38:20,640 --> 00:38:24,000
Speaker 1: Why because an AI can make the decision about shooting

815
00:38:24,120 --> 00:38:26,480
Speaker 1: much faster than a human can. So which drone is

816
00:38:26,520 --> 00:38:26,920
Speaker 1: going to win?

817
00:38:27,200 --> 00:38:30,720
Speaker 3: An AI will be a better fighter than a human fighter.

818
00:38:30,520 --> 00:38:35,160
Speaker 1: Yes, And so eventually these AI will be making kill decisions,

819
00:38:35,239 --> 00:38:37,560
Speaker 1: right because the one that can make the decision faster

820
00:38:37,719 --> 00:38:39,879
Speaker 1: is going to be the one that wins. And so

821
00:38:40,160 --> 00:38:41,880
Speaker 1: I don't think it's going to be very long before

822
00:38:41,920 --> 00:38:47,080
Speaker 1: we have AI powered drones that are authorized to kill people. Right.

823
00:38:47,120 --> 00:38:49,359
Speaker 1: This is a clear next step for the military, you know,

824
00:38:49,480 --> 00:38:51,960
Speaker 1: like here, here's a picture of somebody we think is

825
00:38:51,960 --> 00:38:54,960
Speaker 1: a terrorist. If you spot them, just fire the missile.

826
00:38:54,960 --> 00:38:58,000
Speaker 1: Don't bother checking, Yeah, don't bother checking with us? Right, Oh,

827
00:38:58,680 --> 00:39:01,520
Speaker 1: that's a clear next step. So now you have AI

828
00:39:01,680 --> 00:39:04,080
Speaker 1: that have the authority to kill people, and why because

829
00:39:04,320 --> 00:39:07,080
Speaker 1: they've been tasked to, you know, take care of us

830
00:39:07,160 --> 00:39:08,000
Speaker 1: or protect us or.

831
00:39:08,280 --> 00:39:10,759
Speaker 3: Only only if you give it that permission though, right,

832
00:39:11,239 --> 00:39:13,520
Speaker 3: I mean that's a big ethical step to say, like,

833
00:39:13,640 --> 00:39:14,799
Speaker 3: if you see them, shoot them.

834
00:39:14,880 --> 00:39:16,919
Speaker 1: Yeah, But I don't think that's a big ethical step

835
00:39:16,960 --> 00:39:19,959
Speaker 1: for the military. You know, the protocols for shooting somebody

836
00:39:19,960 --> 00:39:21,799
Speaker 1: in the military. I mean, I'm not an expert on

837
00:39:21,880 --> 00:39:25,680
Speaker 1: military protocols, but you know, our military kills a lot

838
00:39:25,719 --> 00:39:29,719
Speaker 1: of people for you know, a lot of civilians get killed, right,

839
00:39:29,760 --> 00:39:32,360
Speaker 1: and we decide it's okay. A lot of innocent people

840
00:39:32,400 --> 00:39:34,759
Speaker 1: get get killed for military purposes, and so I don't

841
00:39:34,760 --> 00:39:37,840
Speaker 1: think it's too far before AI is making that decision.

842
00:39:39,280 --> 00:39:42,719
Speaker 1: And then it's AI. It's weaponized AI, our weaponized AI

843
00:39:42,880 --> 00:39:45,799
Speaker 1: versus their weaponized AI. And then it's an arms race,

844
00:39:46,120 --> 00:39:48,680
Speaker 1: and then the most powerful army is going to be

845
00:39:48,719 --> 00:39:50,440
Speaker 1: the one that just makes it all of us decisions,

846
00:39:50,480 --> 00:39:54,640
Speaker 1: and the generals just say, defend us right or respond

847
00:39:54,680 --> 00:39:56,920
Speaker 1: if we're attacked right, And then you basically hand it

848
00:39:56,960 --> 00:40:00,239
Speaker 1: over control of the weapons to the AI because the

849
00:40:00,400 --> 00:40:02,000
Speaker 1: enemy has weaponized AI.

850
00:40:02,160 --> 00:40:04,719
Speaker 3: But that doesn't mean that they're controlling us. I mean,

851
00:40:04,760 --> 00:40:07,360
Speaker 3: we use them to protect us or to take away

852
00:40:07,440 --> 00:40:09,359
Speaker 3: some decision making for it, but that doesn't mean that

853
00:40:09,360 --> 00:40:11,680
Speaker 3: they're necessarily in control of us.

854
00:40:11,920 --> 00:40:13,799
Speaker 1: And let's make sure not to be too alarmist here,

855
00:40:13,840 --> 00:40:16,120
Speaker 1: of course, because people are working really hard to make

856
00:40:16,160 --> 00:40:19,000
Speaker 1: sure that there are always ways for humans to override

857
00:40:19,040 --> 00:40:19,680
Speaker 1: these systems.

858
00:40:19,719 --> 00:40:22,560
Speaker 3: Well, it'd be different, that'd be you know, it'd be

859
00:40:22,600 --> 00:40:25,080
Speaker 3: like if a robot then turns the weapons inwards. That's

860
00:40:25,120 --> 00:40:26,719
Speaker 3: another deal, I guess.

861
00:40:27,080 --> 00:40:29,440
Speaker 1: Yeah. And of course AI researchers do their best to

862
00:40:29,480 --> 00:40:31,920
Speaker 1: make sure that the AI systems are very well trained

863
00:40:32,280 --> 00:40:34,839
Speaker 1: so that they do exactly what we want them to do.

864
00:40:35,040 --> 00:40:38,279
Speaker 1: But they are complex and unpredictable, just like people are.

865
00:40:38,360 --> 00:40:43,520
Speaker 3: Right, So this is a very interesting topic, whether AI

866
00:40:43,600 --> 00:40:46,919
Speaker 3: is dangerous or not. And I know, Daniel that you're

867
00:40:47,080 --> 00:40:49,759
Speaker 3: sort of an expert in artificial intelligence because you use

868
00:40:49,800 --> 00:40:54,240
Speaker 3: it in your particle physics research, right, you use machine learning.

869
00:40:54,160 --> 00:40:54,440
Speaker 6: That's right.

870
00:40:54,440 --> 00:40:56,000
Speaker 1: I wouldn't say I'm an expert. I mean I know

871
00:40:56,120 --> 00:40:58,880
Speaker 1: something about it. I've done some reading and I've used it,

872
00:40:58,920 --> 00:41:01,800
Speaker 1: but I'm certainly not at deep expert in artificial intelligence.

873
00:41:01,840 --> 00:41:05,799
Speaker 3: Itself, right, But you know some experts in your department, right,

874
00:41:05,960 --> 00:41:07,240
Speaker 3: and you're in your campus.

875
00:41:06,920 --> 00:41:09,520
Speaker 1: That's right. You see, I has an amazing computer science

876
00:41:09,560 --> 00:41:12,200
Speaker 1: department and experts in machine learning. Some of the folks

877
00:41:12,280 --> 00:41:15,320
Speaker 1: I actually collaborate with. When we're understanding the huge amounts

878
00:41:15,360 --> 00:41:18,680
Speaker 1: of data from the large Hagon collider, we train machines

879
00:41:18,719 --> 00:41:20,720
Speaker 1: to sift through that data and look for the Higgs

880
00:41:20,760 --> 00:41:23,719
Speaker 1: boson and learn to recognize new kinds of particles. It's

881
00:41:23,800 --> 00:41:26,360
Speaker 1: really fun. And these guys know a lot about artificial

882
00:41:26,360 --> 00:41:28,279
Speaker 1: intelligence more than I do. So I went over there

883
00:41:28,320 --> 00:41:30,959
Speaker 1: and I asked them if they were worried about whether

884
00:41:31,239 --> 00:41:32,400
Speaker 1: robots would take over.

885
00:41:32,200 --> 00:41:34,400
Speaker 3: The world, and what did the robots say.

886
00:41:35,800 --> 00:41:38,080
Speaker 1: The robots had taken over the professors and the answers

887
00:41:38,120 --> 00:41:42,759
Speaker 1: form no. First, here's professor Pierre Baldi, he's a distinguished

888
00:41:42,920 --> 00:41:45,440
Speaker 1: professor on campus, and here's what he had to say.

889
00:41:45,719 --> 00:41:51,080
Speaker 10: Potentially, Yes, all very powerful technologies I think can pose

890
00:41:51,200 --> 00:41:54,440
Speaker 10: such a threat, and all depends how they are deployed,

891
00:41:54,560 --> 00:41:57,279
Speaker 10: how they are used, et cetera. Right, you can say

892
00:41:57,280 --> 00:42:01,520
Speaker 10: that nuclear technology poses such a threat and continues to

893
00:42:01,600 --> 00:42:06,480
Speaker 10: pose such a threat, and I think AI, if used

894
00:42:06,480 --> 00:42:10,960
Speaker 10: in the wrong way to pose a threat to mankind.

895
00:42:11,080 --> 00:42:12,960
Speaker 2: Yes, the potential is there.

896
00:42:12,840 --> 00:42:16,319
Speaker 1: And so we should be careful, right, So that was

897
00:42:16,440 --> 00:42:18,799
Speaker 1: Professor Baldy, and then I also went down the hall

898
00:42:18,840 --> 00:42:20,600
Speaker 1: and I asked another colleague. So I thought, let's get

899
00:42:20,600 --> 00:42:23,800
Speaker 1: more than one opinion, and so this is Professor pork Smith,

900
00:42:23,920 --> 00:42:26,840
Speaker 1: also a professor of computer science at UC Irvine.

901
00:42:27,440 --> 00:42:30,600
Speaker 11: I think the main threat with artificial intelligence going forward

902
00:42:30,920 --> 00:42:35,480
Speaker 11: is not understanding how the black boxes work. And so

903
00:42:35,520 --> 00:42:39,160
Speaker 11: I think not the typical sort of we're going to

904
00:42:39,200 --> 00:42:42,520
Speaker 11: have robots taking over the world, but more the use

905
00:42:42,520 --> 00:42:46,239
Speaker 11: of AI in situations where we're extrapolating beyond what it

906
00:42:46,280 --> 00:42:48,560
Speaker 11: can do, and so I think we need to understand

907
00:42:48,560 --> 00:42:49,879
Speaker 11: the limits of AI.

908
00:42:49,960 --> 00:42:52,480
Speaker 1: I think that's a threat, all right, So the answer

909
00:42:52,560 --> 00:42:56,880
Speaker 1: is yes, Well, I think they're cautious, right, both of

910
00:42:56,920 --> 00:42:59,800
Speaker 1: them think it's unpredictable. We don't know what's going to happen.

911
00:43:00,040 --> 00:43:02,400
Speaker 1: We're creating a whole new kind of system, right, and

912
00:43:03,080 --> 00:43:05,319
Speaker 1: we may lose control of parts of it. On the

913
00:43:05,320 --> 00:43:07,600
Speaker 1: other hand, you know, is it likely for that to happen.

914
00:43:08,000 --> 00:43:09,759
Speaker 1: You know, a lot of people are working really hard

915
00:43:09,800 --> 00:43:12,200
Speaker 1: to make sure that AI will be contained and that

916
00:43:12,400 --> 00:43:14,000
Speaker 1: in the end, you could just pull the plug if

917
00:43:14,040 --> 00:43:18,720
Speaker 1: the robot revolution starts, and so it is unpredictable. But also,

918
00:43:18,840 --> 00:43:20,719
Speaker 1: you know, the future is unpredictable, is always going to

919
00:43:20,719 --> 00:43:21,400
Speaker 1: be unpredictable.

920
00:43:21,440 --> 00:43:24,000
Speaker 3: Yeah, I feel like I thought it was interesting. He

921
00:43:24,040 --> 00:43:27,520
Speaker 3: said it is dangerous, but not more so than any

922
00:43:27,600 --> 00:43:29,440
Speaker 3: other powerful technology.

923
00:43:29,760 --> 00:43:32,200
Speaker 1: Yeah, that's a really interesting comment. It's true that any

924
00:43:32,239 --> 00:43:35,120
Speaker 1: technology you can create could be used for good or for.

925
00:43:35,120 --> 00:43:38,160
Speaker 3: Either if it's powerful, like I mean, not just like

926
00:43:38,320 --> 00:43:41,160
Speaker 3: you know, wind up toy. Maybe it's not as dangerous.

927
00:43:41,160 --> 00:43:44,279
Speaker 3: But I think that speaks to the kind of the

928
00:43:44,320 --> 00:43:48,240
Speaker 3: power of AI, Like it really is maybe more powerful

929
00:43:48,280 --> 00:43:49,320
Speaker 3: than we can handle.

930
00:43:49,719 --> 00:43:52,360
Speaker 1: Yeah, and it's powerful in a special way, right, like

931
00:43:52,440 --> 00:43:55,440
Speaker 1: nuclear weapons are powerful. Right, But in the end, a

932
00:43:55,520 --> 00:43:58,040
Speaker 1: human is making that decision, and so you're giving humans

933
00:43:58,040 --> 00:44:00,680
Speaker 1: a new kind of power which is unpredictable. But here

934
00:44:00,760 --> 00:44:04,439
Speaker 1: you're you're unleashing something, right, You're creating AI, and it's

935
00:44:04,480 --> 00:44:07,359
Speaker 1: making its own decisions. Of course, it's making decisions based

936
00:44:07,400 --> 00:44:09,000
Speaker 1: on what has been told to do. Right, you have

937
00:44:09,080 --> 00:44:11,120
Speaker 1: to give an instructions. Still, you have to teach it,

938
00:44:11,800 --> 00:44:14,279
Speaker 1: but you can't predict what these complex systems are going

939
00:44:14,360 --> 00:44:16,160
Speaker 1: to do in new circumstances and how they're going to

940
00:44:16,200 --> 00:44:18,680
Speaker 1: interpret your instructions, right, And of course there are a

941
00:44:18,680 --> 00:44:20,719
Speaker 1: lot of AI smart people out there working hard to

942
00:44:20,719 --> 00:44:24,239
Speaker 1: make sure that there are boundaries and safeties installed in

943
00:44:24,280 --> 00:44:27,120
Speaker 1: all AI systems. But you know, I've seen Jurassic Park,

944
00:44:27,920 --> 00:44:30,359
Speaker 1: you know, the lesson there, they had fences, the lesson there,

945
00:44:30,560 --> 00:44:32,160
Speaker 1: they had fences. We haves.

946
00:44:32,200 --> 00:44:35,920
Speaker 3: So then Jeff Goldboom, you know, has a theory about chaos.

947
00:44:36,120 --> 00:44:38,720
Speaker 1: Yeah, exactly. You know, these systems are hard to predict,

948
00:44:38,760 --> 00:44:41,239
Speaker 1: and so I think we should be worried, but then

949
00:44:41,280 --> 00:44:44,120
Speaker 1: we should respond to that worry with appropriate safeguards, you know,

950
00:44:44,160 --> 00:44:47,360
Speaker 1: and we should take this seriously but not be overly alarmed.

951
00:44:47,440 --> 00:44:47,600
Speaker 7: Right.

952
00:44:47,680 --> 00:44:49,560
Speaker 3: Well, the other point that the other professor made is

953
00:44:49,680 --> 00:44:53,399
Speaker 3: also interesting that he's saying some of the danger is

954
00:44:53,760 --> 00:44:56,680
Speaker 3: in the fact that it's kind of like a black box,

955
00:44:56,920 --> 00:44:59,920
Speaker 3: like we're trusting these things, but we don't really know

956
00:45:00,160 --> 00:45:03,360
Speaker 3: what's going on inside. Like it's so complex that we

957
00:45:04,040 --> 00:45:07,840
Speaker 3: can predict what it's going to do, we can't maybe

958
00:45:07,960 --> 00:45:10,480
Speaker 3: even deconstruct how it makes decisions.

959
00:45:10,719 --> 00:45:13,600
Speaker 1: That's right, And you know, you train these systems, they're

960
00:45:13,680 --> 00:45:15,440
Speaker 1: very complicated and you don't know how they're going to

961
00:45:15,480 --> 00:45:17,040
Speaker 1: respond to new circumstances.

962
00:45:17,120 --> 00:45:17,239
Speaker 4: Right.

963
00:45:17,280 --> 00:45:19,719
Speaker 1: It's the same as like training your dog, Like do

964
00:45:19,760 --> 00:45:21,680
Speaker 1: you know how your dog makes a decision about who

965
00:45:21,680 --> 00:45:23,840
Speaker 1: debarka and who not to. You try to train it,

966
00:45:24,080 --> 00:45:26,040
Speaker 1: You try to give it instructions, try to make sure

967
00:45:26,080 --> 00:45:27,759
Speaker 1: it knows how to how to handle it stuff in

968
00:45:27,760 --> 00:45:30,120
Speaker 1: a new circumstance, but you can't honestly know what it's

969
00:45:30,160 --> 00:45:31,360
Speaker 1: going to do at any given moment.

970
00:45:32,000 --> 00:45:34,200
Speaker 3: Yeah, yeah, I'm definitely not visiting your house if you

971
00:45:34,239 --> 00:45:34,760
Speaker 3: have dogs.

972
00:45:36,920 --> 00:45:40,160
Speaker 1: I think about I think about AI, and again not

973
00:45:40,200 --> 00:45:43,160
Speaker 1: an expert, so maybe these are uninformed speculations, but I

974
00:45:43,200 --> 00:45:46,279
Speaker 1: think about AI sort of like digital children. You know,

975
00:45:46,400 --> 00:45:48,680
Speaker 1: like you raise your children, you know they're going to

976
00:45:48,719 --> 00:45:50,400
Speaker 1: take over one day because you know you and I

977
00:45:50,440 --> 00:45:52,120
Speaker 1: are going to get old and our kids are younger

978
00:45:52,120 --> 00:45:54,919
Speaker 1: than we are, so eventually they will take over and

979
00:45:55,320 --> 00:45:56,960
Speaker 1: you don't know what they're going to do, and you

980
00:45:57,040 --> 00:45:58,759
Speaker 1: raise them. You try to raise them in a way

981
00:45:58,760 --> 00:46:01,799
Speaker 1: that they have values, they make reasonable decisions, And you

982
00:46:01,840 --> 00:46:03,839
Speaker 1: can sort of think about AI the same way, like

983
00:46:04,400 --> 00:46:07,600
Speaker 1: you try to create this new generation of technology that's

984
00:46:07,600 --> 00:46:09,160
Speaker 1: going to make its own decisions, but you try to

985
00:46:09,200 --> 00:46:11,560
Speaker 1: teach it to make good decisions so that when you're

986
00:46:11,560 --> 00:46:14,120
Speaker 1: in a home, right, it's making good choices for you.

987
00:46:14,480 --> 00:46:16,480
Speaker 1: And I know that some folks out there think, well,

988
00:46:16,520 --> 00:46:18,240
Speaker 1: you know, AI is never really going to be separate

989
00:46:18,239 --> 00:46:21,640
Speaker 1: from humanity. It's not this like cognitive separation, like you

990
00:46:21,680 --> 00:46:23,399
Speaker 1: can just be part of who you are, the way

991
00:46:23,400 --> 00:46:26,319
Speaker 1: your iPhone feels like part of who you are. But

992
00:46:26,400 --> 00:46:29,560
Speaker 1: we don't know necessarily if if that separation is going

993
00:46:29,600 --> 00:46:31,680
Speaker 1: to be serious, you know, if these things really will

994
00:46:31,760 --> 00:46:33,359
Speaker 1: be separate from us, or if they always just feel

995
00:46:33,360 --> 00:46:34,680
Speaker 1: like an extension of ourselves.

996
00:46:35,680 --> 00:46:40,760
Speaker 3: Well, until then, I think we should stick to regular dogs. Dogs.

997
00:46:42,880 --> 00:46:44,680
Speaker 1: Yeah, But you know, I think about it sometimes the

998
00:46:44,719 --> 00:46:47,520
Speaker 1: way I think about children, right, in the same way

999
00:46:47,560 --> 00:46:50,319
Speaker 1: that you raise your children and they're going to take over, right,

1000
00:46:50,320 --> 00:46:52,440
Speaker 1: there's gonna be some point when your children are in charge.

1001
00:46:52,800 --> 00:46:55,480
Speaker 1: You raise them to have values and to make good decisions,

1002
00:46:55,480 --> 00:46:57,759
Speaker 1: and you hope that when they take over, they're you know,

1003
00:46:57,880 --> 00:47:00,759
Speaker 1: looking after you in the same way. To create these

1004
00:47:00,800 --> 00:47:03,400
Speaker 1: digital tools, and we got to teach them to behave

1005
00:47:03,400 --> 00:47:05,080
Speaker 1: We got to teach them what's important, and we got

1006
00:47:05,080 --> 00:47:07,879
Speaker 1: to teach them how to be responsible so that if

1007
00:47:07,920 --> 00:47:10,480
Speaker 1: they take over, you know, that we hope they treat

1008
00:47:10,520 --> 00:47:10,919
Speaker 1: as well.

1009
00:47:11,080 --> 00:47:15,440
Speaker 3: Yeah, daddy good, your parents.

1010
00:47:15,200 --> 00:47:20,120
Speaker 1: Don't put creator home, Please don't bury me underground.

1011
00:47:25,600 --> 00:47:27,719
Speaker 3: Well, I personally am looking forward to a time when

1012
00:47:27,760 --> 00:47:29,680
Speaker 3: I have, like I don't have to think as much,

1013
00:47:30,000 --> 00:47:32,480
Speaker 3: where life is a little bit easier because we have

1014
00:47:32,640 --> 00:47:35,920
Speaker 3: these things making things easier for us.

1015
00:47:36,040 --> 00:47:38,040
Speaker 1: It could handle a lot of the drudgery and a

1016
00:47:38,080 --> 00:47:40,400
Speaker 1: lot of the logistics. You know, eventually you could have

1017
00:47:40,400 --> 00:47:43,200
Speaker 1: a car that drives itself and obeys your instructions. You

1018
00:47:43,239 --> 00:47:45,200
Speaker 1: could say, like, hey, go pick up my kids from school,

1019
00:47:45,520 --> 00:47:47,279
Speaker 1: and you would know how to navigate and how to

1020
00:47:47,360 --> 00:47:50,520
Speaker 1: drive and recognize your children and how to get back home.

1021
00:47:50,560 --> 00:47:53,160
Speaker 1: And that's totally within the realm of possibility in a

1022
00:47:53,200 --> 00:47:55,800
Speaker 1: few years, right, And that's pretty awesome. It'll offload a

1023
00:47:55,840 --> 00:47:58,520
Speaker 1: lot of work and logistics from belieguered parents.

1024
00:47:58,600 --> 00:48:00,400
Speaker 3: I think you and I Aron are pretty good position

1025
00:48:00,560 --> 00:48:03,399
Speaker 3: career wise, you know, like I'm a cartoonist and you're

1026
00:48:03,400 --> 00:48:06,759
Speaker 3: a physicist. These are not jobs that are going to

1027
00:48:06,840 --> 00:48:09,080
Speaker 3: be taken away by AI anytime soon.

1028
00:48:09,280 --> 00:48:12,319
Speaker 1: Hopefully have you not seen AI cartoons? They're pretty good, man.

1029
00:48:14,160 --> 00:48:16,560
Speaker 1: You should like start a podcast instead of wearing instead

1030
00:48:16,560 --> 00:48:17,719
Speaker 1: of relying on your cartooning.

1031
00:48:18,239 --> 00:48:21,239
Speaker 3: Well, there is definitely that as a genre of humor, Like, hey,

1032
00:48:21,280 --> 00:48:24,200
Speaker 3: I put so and so through an AI machine and

1033
00:48:24,200 --> 00:48:26,040
Speaker 3: look at the crazy thing it came out with.

1034
00:48:26,640 --> 00:48:29,160
Speaker 1: Except those are all manufactured. None of those are none

1035
00:48:29,160 --> 00:48:32,560
Speaker 1: of those are real, None of those are real, those

1036
00:48:32,560 --> 00:48:33,319
Speaker 1: are all made up.

1037
00:48:33,440 --> 00:48:35,759
Speaker 3: Well, that's good for humors.

1038
00:48:36,280 --> 00:48:39,680
Speaker 1: So artificial intelligence is certainly a revolution in thinking and

1039
00:48:39,719 --> 00:48:42,560
Speaker 1: in computing, and it will definitely change the world. And

1040
00:48:42,640 --> 00:48:44,839
Speaker 1: so check back in in ten years to see if

1041
00:48:44,880 --> 00:48:48,000
Speaker 1: we've been replaced by robot Daniel and robot Horge.

1042
00:48:48,280 --> 00:48:50,800
Speaker 3: Maybe we already are bump bump bomb.

1043
00:48:51,560 --> 00:48:53,959
Speaker 1: So thanks everyone for listening to this episode of Daniel

1044
00:48:54,000 --> 00:48:55,600
Speaker 1: and Jorge Explain the Universe.

1045
00:48:55,719 --> 00:48:59,359
Speaker 3: Yeah, and to listen to more, just say, Alexa, what's

1046
00:48:59,400 --> 00:49:02,520
Speaker 3: the best sign this podcast in the world? Well, what's

1047
00:49:02,520 --> 00:49:04,960
Speaker 3: the third best podcast in the world.

1048
00:49:13,600 --> 00:49:15,879
Speaker 1: If you still have a question after listening to all

1049
00:49:15,920 --> 00:49:19,160
Speaker 1: these explanations, please drop us a line. We'd love to

1050
00:49:19,160 --> 00:49:21,640
Speaker 1: hear from you. You can find us at Facebook, Twitter,

1051
00:49:21,680 --> 00:49:25,360
Speaker 1: and Instagram at Daniel and Jorge That's one word, or

1052
00:49:25,480 --> 00:49:38,160
Speaker 1: email us at Feedback at Danielanorge dot com. When you

1053
00:49:38,200 --> 00:49:40,240
Speaker 1: pop a piece of cheese into your mouth, You're probably

1054
00:49:40,320 --> 00:49:43,360
Speaker 1: not thinking about the environmental impact, but the people in

1055
00:49:43,400 --> 00:49:46,520
Speaker 1: the dairy industry are. That's why they're working hard every

1056
00:49:46,600 --> 00:49:49,920
Speaker 1: day to find new ways to reduce waste, conserve natural resources,

1057
00:49:49,920 --> 00:49:53,520
Speaker 1: and drive down greenhouse gas emissions. How is us dairy

1058
00:49:53,560 --> 00:49:57,680
Speaker 1: tackling greenhouse gases? Many farms use anaerobic digestors to turn

1059
00:49:57,719 --> 00:50:02,200
Speaker 1: the methane from manure into renewable energy that can power farms, towns,

1060
00:50:02,239 --> 00:50:05,520
Speaker 1: and electric cars. Visit you as dairy dot COM's Last

1061
00:50:05,520 --> 00:50:07,160
Speaker 1: Sustainability to learn more.

1062
00:50:07,920 --> 00:50:12,200
Speaker 12: This is Malcolm Gladwell from Revisionist History. eBay Motors is

1063
00:50:12,280 --> 00:50:16,279
Speaker 12: here for the ride. With samelbow grease, fresh installs, and

1064
00:50:16,320 --> 00:50:19,040
Speaker 12: a whole lot of love, you transformed one hundred thousand

1065
00:50:19,040 --> 00:50:21,600
Speaker 12: miles and a body full of rust into a drive

1066
00:50:21,640 --> 00:50:25,799
Speaker 12: that's all your own, break kits, led headlights, whatever you need.

1067
00:50:26,120 --> 00:50:30,560
Speaker 12: eBay Motors has it and with eBay Guaranteed Fit, it's

1068
00:50:30,640 --> 00:50:34,640
Speaker 12: guaranteed to fit your ride the first time, every time,

1069
00:50:35,239 --> 00:50:39,200
Speaker 12: or your money back plus. At these prices, you're burning rubber,

1070
00:50:39,239 --> 00:50:42,880
Speaker 12: not cash. Keep your ride or die alive at ebaymotors

1071
00:50:42,880 --> 00:50:46,480
Speaker 12: dot Com. Eligible items only. Exclusions apply.

1072
00:50:46,719 --> 00:50:49,320
Speaker 6: As a United Explorer Card member, you can earn fifty

1073
00:50:49,320 --> 00:50:53,240
Speaker 6: thousand bonus miles plus look forward to extraordinary travel rewards,

1074
00:50:53,320 --> 00:50:55,840
Speaker 6: including a free checked bag, two times the miles on

1075
00:50:55,920 --> 00:50:58,600
Speaker 6: United purchases and two times the miles on dining and

1076
00:50:58,640 --> 00:51:01,960
Speaker 6: at hotels. Become an explore and seek out unforgettable places

1077
00:51:01,960 --> 00:51:05,359
Speaker 6: while enjoying rewards everywhere you travel. Cards issued by JP

1078
00:51:05,480 --> 00:51:09,120
Speaker 6: morgan Chase Bank NA Member FDIC subject to credit approval

1079
00:51:09,239 --> 00:51:10,480
Speaker 6: Offer subject to change.

1080
00:51:10,640 --> 00:51:11,360
Speaker 3: Terms apply