1
00:00:00,200 --> 00:00:02,880
Speaker 1: Thanks for tunion to tech Stuff. If you don't recognize

2
00:00:02,880 --> 00:00:06,000
Speaker 1: my voice, my name is Osvoloshan, and I'm here because

3
00:00:06,040 --> 00:00:09,160
Speaker 1: the inimitable Jonathan Strickland has passed the baton to Kara

4
00:00:09,240 --> 00:00:12,239
Speaker 1: Price and myself to host tech Stuff. The show will

5
00:00:12,240 --> 00:00:15,000
Speaker 1: remain your home for all things tech, and all the

6
00:00:15,040 --> 00:00:18,759
Speaker 1: old episodes will remain available in this feed. Thanks for listening.

7
00:00:20,520 --> 00:00:24,520
Speaker 1: Welcome to Tech Stuff, a production from iHeartRadio and Kaleidoscope.

8
00:00:25,000 --> 00:00:28,840
Speaker 1: I'm Osvoloshian today co host Kara Price, and I will

9
00:00:28,840 --> 00:00:33,360
Speaker 1: bring you three things. First, the headlines this week. Second,

10
00:00:33,760 --> 00:00:37,000
Speaker 1: a conversation with four O four Media's Jason Kebler about

11
00:00:37,040 --> 00:00:41,280
Speaker 1: some deceptive AI bots looking to trick their human counterparts

12
00:00:41,400 --> 00:00:44,360
Speaker 1: in today's Tech Support segment. And finally, we head to

13
00:00:44,400 --> 00:00:47,400
Speaker 1: see yes kind of. We take a look back at

14
00:00:47,400 --> 00:00:50,040
Speaker 1: the year in tech with Emma Barker of Time magazine,

15
00:00:50,240 --> 00:00:53,160
Speaker 1: who edited Time's list of the two hundred best inventions

16
00:00:53,159 --> 00:00:55,800
Speaker 1: of twenty twenty four. All of that on this Week

17
00:00:55,840 --> 00:01:19,280
Speaker 1: in Tech. It's Friday, January tenth. Stay with us, so Carat,

18
00:01:19,280 --> 00:01:21,800
Speaker 1: it's very nice to see you. It's been almost half

19
00:01:21,840 --> 00:01:24,600
Speaker 1: a decade since we've been in studio together. A pandemic,

20
00:01:24,760 --> 00:01:25,960
Speaker 1: A pandemic.

21
00:01:25,720 --> 00:01:28,679
Speaker 2: Many developments in tech, very few in my own personal life.

22
00:01:29,840 --> 00:01:31,960
Speaker 2: But you know, there's nothing like a big mic in

23
00:01:31,959 --> 00:01:34,639
Speaker 2: my face to make me feel like a normal person again.

24
00:01:34,920 --> 00:01:37,440
Speaker 2: And I know we thank Jonathan already in our last

25
00:01:37,480 --> 00:01:41,240
Speaker 2: episode for bringing us back together, but I want to

26
00:01:41,240 --> 00:01:43,160
Speaker 2: get him a gift, and I think it's going to

27
00:01:43,200 --> 00:01:45,680
Speaker 2: be to see the show. Oh, Mary, I didn't know

28
00:01:45,720 --> 00:01:47,320
Speaker 2: he was such a theater buff.

29
00:01:47,640 --> 00:01:49,520
Speaker 1: I remember because we did an interview with him a

30
00:01:49,520 --> 00:01:52,800
Speaker 1: few years ago that he was a big Shakespeare guy. Oh.

31
00:01:52,880 --> 00:01:53,160
Speaker 2: Yes.

32
00:01:53,400 --> 00:01:55,520
Speaker 1: He seemed to think it would be easier for us

33
00:01:55,560 --> 00:01:58,000
Speaker 1: to host the show as two people or rather than

34
00:01:58,040 --> 00:01:59,800
Speaker 1: as just him, and I kind of see his point,

35
00:01:59,800 --> 00:02:02,200
Speaker 1: But I also think he probably hasn't spent quite enough

36
00:02:02,240 --> 00:02:05,120
Speaker 1: time with either of us to make a truly informed

37
00:02:06,080 --> 00:02:07,200
Speaker 1: speculation about that.

38
00:02:07,840 --> 00:02:10,160
Speaker 2: You know, I think the reason that this show works

39
00:02:10,200 --> 00:02:12,120
Speaker 2: between the two of us is I think you and

40
00:02:12,200 --> 00:02:17,040
Speaker 2: I sort of speak to each other in link tongues.

41
00:02:17,120 --> 00:02:19,520
Speaker 2: I always say, you know, there's the love languages. Yeah,

42
00:02:19,960 --> 00:02:23,480
Speaker 2: I have a ace love language, which is links links, and.

43
00:02:23,400 --> 00:02:26,400
Speaker 1: That's yeah tongues. Okay. I was going to say, actually,

44
00:02:26,440 --> 00:02:31,160
Speaker 1: for the avoidance of doubt, just friends.

45
00:02:31,320 --> 00:02:35,040
Speaker 2: We always say as Sherlock and Watson the most platonic

46
00:02:35,080 --> 00:02:37,079
Speaker 2: and they live together on Baker's story exactly.

47
00:02:37,560 --> 00:02:39,160
Speaker 1: So so no, I mean it's going to be It's

48
00:02:39,160 --> 00:02:40,680
Speaker 1: going to be the two of us with two people.

49
00:02:40,720 --> 00:02:43,320
Speaker 1: But we are in constant communication about all things tech

50
00:02:43,639 --> 00:02:46,280
Speaker 1: and hopefully we'll be able to shine a bit of

51
00:02:46,280 --> 00:02:48,720
Speaker 1: a light on some of the most interesting intriguing things

52
00:02:48,800 --> 00:02:52,160
Speaker 1: each week, including this one. So, Caara, we've been texting

53
00:02:52,200 --> 00:02:53,680
Speaker 1: a lot. What have you been reading that really stood

54
00:02:53,720 --> 00:02:54,040
Speaker 1: out to you?

55
00:02:54,680 --> 00:02:54,760
Speaker 3: So?

56
00:02:55,000 --> 00:02:58,160
Speaker 2: I think, like everyone in America at least, I've been

57
00:02:58,200 --> 00:03:02,720
Speaker 2: following the cyber truck explosion in Las Vegas and Futurism,

58
00:03:02,840 --> 00:03:05,120
Speaker 2: which is a website I frequent picked up on some

59
00:03:05,240 --> 00:03:09,040
Speaker 2: original reporting from the Associated Press that has been kind

60
00:03:09,080 --> 00:03:14,880
Speaker 2: of quiet. The bomber YEP, who's an active duty Green Beret. Yeah,

61
00:03:15,440 --> 00:03:17,120
Speaker 2: used chat GPT.

62
00:03:17,880 --> 00:03:19,600
Speaker 1: Do you know what I saw this this morning? I

63
00:03:19,639 --> 00:03:21,720
Speaker 1: was so sad this wasn't on my list. I'm happy

64
00:03:21,800 --> 00:03:22,680
Speaker 1: this on yours.

65
00:03:23,080 --> 00:03:25,760
Speaker 2: Like, I don't mean to make light of the situation

66
00:03:25,800 --> 00:03:28,840
Speaker 2: because it's it's very depressing on a number of levels,

67
00:03:29,320 --> 00:03:31,040
Speaker 2: but I knew you would see the story.

68
00:03:31,600 --> 00:03:33,760
Speaker 1: No, I was fascinated, but It was the fourth story

69
00:03:33,800 --> 00:03:36,680
Speaker 1: on Exios today, and I'm glad that it's the first

70
00:03:36,720 --> 00:03:39,120
Speaker 1: story on your show because I found it totally mind blowing.

71
00:03:39,280 --> 00:03:43,320
Speaker 2: And obviously every part of this story is terrible, including

72
00:03:43,360 --> 00:03:45,600
Speaker 2: the fact that a cyber truck was involved. Anything involving

73
00:03:45,600 --> 00:03:47,760
Speaker 2: a cyber truck makes me want to die. Yeah, I

74
00:03:47,800 --> 00:03:50,280
Speaker 2: hate that car more than I hate most things. But

75
00:03:50,800 --> 00:03:54,800
Speaker 2: what was so interesting to me about it was it

76
00:03:54,880 --> 00:04:01,200
Speaker 2: marks this moment of really underlying how next generation search

77
00:04:01,960 --> 00:04:07,320
Speaker 2: works YEP on like a very disturbingly practical level. And

78
00:04:07,880 --> 00:04:10,920
Speaker 2: Kevin McHale, who is the Las Vegas County Sheriff, says

79
00:04:10,960 --> 00:04:14,120
Speaker 2: that it's the first known us of chat GPT to

80
00:04:14,200 --> 00:04:17,880
Speaker 2: help build an explosive and he he followed up by saying,

81
00:04:17,960 --> 00:04:19,239
Speaker 2: it's a concerning.

82
00:04:18,800 --> 00:04:20,560
Speaker 1: Moment under statement of the week.

83
00:04:21,040 --> 00:04:23,560
Speaker 2: This is not someone googling.

84
00:04:23,360 --> 00:04:25,600
Speaker 1: The anarchist cookbook or what was that old book.

85
00:04:25,320 --> 00:04:27,360
Speaker 2: That the anarchist ye cookbook.

86
00:04:27,400 --> 00:04:29,360
Speaker 1: I think that was a huge moment, and I think

87
00:04:29,400 --> 00:04:31,600
Speaker 1: the history of publication and First Amendment whether you could

88
00:04:31,680 --> 00:04:33,280
Speaker 1: like distribute a book that told you how to make

89
00:04:33,320 --> 00:04:35,679
Speaker 1: a bomb. But the cat's out of the bag now.

90
00:04:35,720 --> 00:04:37,599
Speaker 2: Well, and I'll tell you who knows it's out of

91
00:04:37,600 --> 00:04:41,520
Speaker 2: the back open ai right, Because for open ai to

92
00:04:41,839 --> 00:04:46,039
Speaker 2: very quickly respond to this, I think is really interesting.

93
00:04:46,080 --> 00:04:50,120
Speaker 1: Did they volunteer? By the way, oops, ps the guy

94
00:04:50,200 --> 00:04:52,600
Speaker 1: used chet gpt to plan this? Or how did it

95
00:04:52,640 --> 00:04:54,440
Speaker 1: come out? Do we know how it came out?

96
00:04:54,680 --> 00:04:58,919
Speaker 2: The Las Vegas Sheriff announced that it was clear that

97
00:04:59,040 --> 00:05:03,039
Speaker 2: chat gbt had been used to build this explosive and

98
00:05:03,120 --> 00:05:06,600
Speaker 2: so open ai then sent an email that was quoted

99
00:05:06,600 --> 00:05:10,359
Speaker 2: in Futurism that said, in this case, chat gpt responded

100
00:05:10,400 --> 00:05:14,720
Speaker 2: with information already publicly available on the internet and provided

101
00:05:14,800 --> 00:05:19,040
Speaker 2: warnings against harmful or illegal activities. We're working with law

102
00:05:19,120 --> 00:05:23,880
Speaker 2: enforcement to support their investigation. And just before you say anything,

103
00:05:24,839 --> 00:05:27,080
Speaker 2: I'm really surprised, like as you were saying, it's a

104
00:05:27,080 --> 00:05:30,520
Speaker 2: four store anaxios, I don't understand how this isn't like

105
00:05:31,080 --> 00:05:32,840
Speaker 2: the lead. I agree, but I guess that's why we're

106
00:05:32,839 --> 00:05:39,000
Speaker 2: doing a tech podcast. So what's what's on your docket today?

107
00:05:39,480 --> 00:05:42,160
Speaker 1: Well, you started with a cyber truck. I've also got

108
00:05:42,520 --> 00:05:45,920
Speaker 1: a vehicle story for you. There was a story in

109
00:05:45,960 --> 00:05:49,800
Speaker 1: the Wall Street Journal over the holidays I found pretty fascinating,

110
00:05:50,279 --> 00:05:55,799
Speaker 1: which is basically a study on Alzheimer's in different populations,

111
00:05:56,240 --> 00:05:59,920
Speaker 1: and the headline was want to avoid Alzheimer's. Taxi drive

112
00:06:00,279 --> 00:06:02,839
Speaker 1: can show you how interesting?

113
00:06:03,720 --> 00:06:05,240
Speaker 2: Say more so?

114
00:06:05,360 --> 00:06:09,480
Speaker 1: Basically, mass General Brigham Hospital in Boston ran a study

115
00:06:09,560 --> 00:06:13,960
Speaker 1: about the rate of Alzheimer's in various populations. Taxi drivers

116
00:06:14,200 --> 00:06:18,920
Speaker 1: and ambulance drivers had up to four times lower rate

117
00:06:18,960 --> 00:06:22,320
Speaker 1: of Alzheimer's than the general population. And apparently this actually

118
00:06:22,360 --> 00:06:25,119
Speaker 1: made sense to researches because the part of the brain

119
00:06:25,200 --> 00:06:30,080
Speaker 1: that does real time spatial processing and decision making, the hippocampus,

120
00:06:30,120 --> 00:06:32,479
Speaker 1: as I learned is called, is also one of the

121
00:06:32,480 --> 00:06:34,480
Speaker 1: first parts of atrophies when you get Alzheimer's.

122
00:06:34,960 --> 00:06:37,479
Speaker 2: That's so interesting. So people who have made this their

123
00:06:37,800 --> 00:06:44,400
Speaker 2: job essentially are people who are kind of saving themselves unknowingly.

124
00:06:44,040 --> 00:06:46,800
Speaker 1: One hundred percent. One of the most interesting things I

125
00:06:46,880 --> 00:06:50,320
Speaker 1: thought was that London bus drivers don't get the same

126
00:06:50,360 --> 00:06:54,120
Speaker 1: benefit as London taxi drivers because they follow a preset route.

127
00:06:54,160 --> 00:06:55,960
Speaker 2: I was going to say that one of my favorite

128
00:06:56,000 --> 00:06:58,599
Speaker 2: things is just the old facts about how much a

129
00:06:58,640 --> 00:07:00,839
Speaker 2: London taxi driver has to know people. What is it

130
00:07:00,880 --> 00:07:01,599
Speaker 2: called the why.

131
00:07:01,560 --> 00:07:04,240
Speaker 1: The knowledge they have when when they when they die

132
00:07:04,279 --> 00:07:07,839
Speaker 1: in the autopsy, they have larger other people. But you

133
00:07:07,880 --> 00:07:10,400
Speaker 1: know who else doesn't get these benefits? Yes, come on

134
00:07:11,080 --> 00:07:15,040
Speaker 1: New Yorkers. People who use Google Maps or ways or

135
00:07:15,080 --> 00:07:17,360
Speaker 1: Apple Maps and what do we call that? We call

136
00:07:17,400 --> 00:07:18,440
Speaker 1: that automation bias.

137
00:07:18,440 --> 00:07:19,520
Speaker 2: Correct, that's our fave.

138
00:07:20,200 --> 00:07:22,160
Speaker 1: So I just found this really interesting. Here. Here is

139
00:07:22,200 --> 00:07:24,800
Speaker 1: the technology Google Maps that I use all the time

140
00:07:25,080 --> 00:07:27,880
Speaker 1: every day, and I absolutely adore it. And I literally

141
00:07:28,000 --> 00:07:30,520
Speaker 1: every time I go on vacation, I think, how the

142
00:07:30,560 --> 00:07:32,400
Speaker 1: hell would I have had a good time if I

143
00:07:32,400 --> 00:07:36,640
Speaker 1: didn't literally literally divorce. Yeah, I would literally go to

144
00:07:36,760 --> 00:07:39,600
Speaker 1: some like the concierge or whatever, or the some random

145
00:07:39,600 --> 00:07:41,880
Speaker 1: person who worked in the hotel say can you give

146
00:07:41,920 --> 00:07:43,480
Speaker 1: me a physical map and tell me how I have

147
00:07:43,520 --> 00:07:45,680
Speaker 1: a good time? I'd be like, I would never travel.

148
00:07:45,720 --> 00:07:48,080
Speaker 2: So, but isn't it what we always talk about, which

149
00:07:48,120 --> 00:07:50,400
Speaker 2: is the sort of the double edge totally?

150
00:07:50,440 --> 00:07:52,480
Speaker 1: And but the idea that using this scene that I

151
00:07:52,520 --> 00:07:56,000
Speaker 1: love every day is you know, shrinking my hippocampus I

152
00:07:56,040 --> 00:07:57,400
Speaker 1: find really quite chilling.

153
00:07:57,640 --> 00:07:59,000
Speaker 2: I mean, I think about this in terms of my

154
00:07:59,360 --> 00:08:02,440
Speaker 2: hands and my Sum's true, Like I just think all

155
00:08:02,480 --> 00:08:04,640
Speaker 2: the time about just burning nerves.

156
00:08:04,800 --> 00:08:07,040
Speaker 1: Well, the devil has no time to make use of

157
00:08:07,080 --> 00:08:09,320
Speaker 1: your sums because they are never idle.

158
00:08:09,920 --> 00:08:13,520
Speaker 2: They're definitely not idle. I think you have one more thing.

159
00:08:13,720 --> 00:08:16,440
Speaker 1: I do have one more thing. And this, don't worry.

160
00:08:16,440 --> 00:08:17,840
Speaker 1: We're not going to talk about that length thing I

161
00:08:17,880 --> 00:08:21,880
Speaker 1: talked to somebody else about. But this was basically, you know,

162
00:08:21,880 --> 00:08:24,840
Speaker 1: there are these new open AI models, one that was

163
00:08:24,920 --> 00:08:28,400
Speaker 1: released in September last year, and three right before the

164
00:08:28,400 --> 00:08:31,760
Speaker 1: holiday on the twelfth day of ship mess as the

165
00:08:31,800 --> 00:08:35,400
Speaker 1: execs called it, but one was being red teamed. Do

166
00:08:35,440 --> 00:08:40,000
Speaker 1: you know what red teaming is. It's when you try

167
00:08:40,080 --> 00:08:43,440
Speaker 1: and make a technological product to break its own rules

168
00:08:43,640 --> 00:08:47,200
Speaker 1: or do unsafe things. So, oh, one was being red teamed,

169
00:08:47,600 --> 00:08:51,440
Speaker 1: and surprise, surprise, in a very interesting way, it started

170
00:08:51,440 --> 00:08:55,120
Speaker 1: trying to deceive its human counterpart. And that's exactly what

171
00:08:55,160 --> 00:08:57,000
Speaker 1: we're going to talk about on our next segment, which

172
00:08:57,040 --> 00:09:01,480
Speaker 1: is Tech Support. Every week we'll do a segment called

173
00:09:01,520 --> 00:09:04,960
Speaker 1: tech Support when we talk to true experts and reporters

174
00:09:05,080 --> 00:09:07,560
Speaker 1: who can go far deeper behind the headlines than you

175
00:09:07,600 --> 00:09:10,080
Speaker 1: and I can, to basically help us sort the signal

176
00:09:10,120 --> 00:09:13,120
Speaker 1: from the noise. And I genuinely don't think there's anyone

177
00:09:13,160 --> 00:09:15,199
Speaker 1: better to talk to than the team at four or

178
00:09:15,280 --> 00:09:18,679
Speaker 1: four Media. These are reporters who were formerly basically the

179
00:09:18,760 --> 00:09:22,000
Speaker 1: team at Vice's motherboard, and now they've started their own collective,

180
00:09:22,600 --> 00:09:25,120
Speaker 1: and they are the people who the tech world follows

181
00:09:25,280 --> 00:09:28,120
Speaker 1: most closely to find out what's really going on in

182
00:09:28,240 --> 00:09:29,960
Speaker 1: all corners of the digital world.

183
00:09:30,559 --> 00:09:33,000
Speaker 2: Yeah, and this week we're excited to share this conversation

184
00:09:33,120 --> 00:09:35,320
Speaker 2: we had with Jason Kebler, who's one of the co

185
00:09:35,440 --> 00:09:38,000
Speaker 2: founders at four h four, who will be filling us

186
00:09:38,040 --> 00:09:41,800
Speaker 2: in on the study that the initial excitement around the

187
00:09:42,040 --> 00:09:46,280
Speaker 2: three release meant did not get enough attention, which revealed

188
00:09:46,280 --> 00:09:50,280
Speaker 2: that cutting edge AI systems, including both open AI's one

189
00:09:50,440 --> 00:09:54,719
Speaker 2: and Anthropics Claude three point five, sonet whatever the way

190
00:09:54,760 --> 00:09:57,360
Speaker 2: they name this stuff is so self agrandizing, have been

191
00:09:57,440 --> 00:10:02,040
Speaker 2: shown to have a shocking t tendency to deceive.

192
00:10:01,840 --> 00:10:07,040
Speaker 1: Shocking tendency to deceive. Indeed, So we talked to Jason

193
00:10:07,120 --> 00:10:09,720
Speaker 1: right before the holidays when this news just came out,

194
00:10:10,040 --> 00:10:10,480
Speaker 1: and we.

195
00:10:10,480 --> 00:10:13,319
Speaker 2: Jump right in, Jason, what's popping.

196
00:10:15,080 --> 00:10:18,000
Speaker 4: So there's a new paper out by Apollo Research, which

197
00:10:18,160 --> 00:10:21,640
Speaker 4: is this group that's funded by various governments as well

198
00:10:21,640 --> 00:10:28,040
Speaker 4: as various artificial intelligence companies like Google, DeepMind, Inthropic, open Ai,

199
00:10:28,160 --> 00:10:30,880
Speaker 4: et cetera. And it's a new paper about what's called

200
00:10:30,960 --> 00:10:37,920
Speaker 4: AI alignment, which is the idea that is an artificial

201
00:10:38,000 --> 00:10:42,880
Speaker 4: intelligence tool doing what the human asked or is it

202
00:10:43,000 --> 00:10:48,280
Speaker 4: doing something else. I think that one of the really

203
00:10:48,679 --> 00:10:53,720
Speaker 4: scary things about artificial intelligence is that they are largely

204
00:10:53,760 --> 00:10:57,600
Speaker 4: black boxes. We're not totally sure how they work, and

205
00:10:57,920 --> 00:11:02,400
Speaker 4: this can lead to all sorts of very interesting outcomes.

206
00:11:02,880 --> 00:11:06,640
Speaker 4: You know, in this case, the researchers at Apollo Research

207
00:11:06,679 --> 00:11:09,520
Speaker 4: Group were able to find that these sort of cutting

208
00:11:09,720 --> 00:11:13,360
Speaker 4: edge large language models, which you know, one of the

209
00:11:13,400 --> 00:11:15,640
Speaker 4: biggest ones is open ais oh one.

210
00:11:15,960 --> 00:11:17,880
Speaker 1: They finally gave it immigrant visa status.

211
00:11:19,720 --> 00:11:22,760
Speaker 4: So they found that in some cases, these sort of

212
00:11:22,920 --> 00:11:27,200
Speaker 4: newest large language models have the I don't even want

213
00:11:27,200 --> 00:11:30,920
Speaker 4: to say ability because I don't want to anthropomorphize them

214
00:11:31,040 --> 00:11:34,679
Speaker 4: so much, but like the tendency to deceive AI is

215
00:11:34,760 --> 00:11:39,239
Speaker 4: quote covertly pursuing goals that are misaligned from its developers

216
00:11:39,320 --> 00:11:43,080
Speaker 4: or users. And examples of this are like, if you

217
00:11:43,240 --> 00:11:47,600
Speaker 4: ask it to answer questions on a test, the artificial

218
00:11:47,600 --> 00:11:50,360
Speaker 4: intelligence would be worried that if it performed too well

219
00:11:50,400 --> 00:11:54,040
Speaker 4: on the test that it would be turned off or deprogrammed,

220
00:11:54,360 --> 00:11:57,600
Speaker 4: and so it would give answers that are not so good,

221
00:11:57,760 --> 00:11:59,600
Speaker 4: so that the human would think that it is not

222
00:11:59,760 --> 00:12:04,120
Speaker 4: as smart as it actually is. In other examples, it

223
00:12:04,760 --> 00:12:10,319
Speaker 4: copied itself onto another server more or less so that

224
00:12:10,440 --> 00:12:13,800
Speaker 4: it was like multiplying itself, which is something that straight

225
00:12:13,800 --> 00:12:17,280
Speaker 4: out of science fiction. I've watched many science fiction movies

226
00:12:17,360 --> 00:12:20,280
Speaker 4: where the artificial intelligence doesn't want to die and so

227
00:12:20,360 --> 00:12:22,640
Speaker 4: it makes like millions of copies of itself all over

228
00:12:22,679 --> 00:12:25,160
Speaker 4: the Internet. And so this type of I mean, I

229
00:12:25,200 --> 00:12:28,319
Speaker 4: guess you call it emergent behavior, which means that it's

230
00:12:28,360 --> 00:12:32,840
Speaker 4: behavior that is not expected and not really programmed in,

231
00:12:33,559 --> 00:12:34,720
Speaker 4: is quite concerning.

232
00:12:35,040 --> 00:12:37,120
Speaker 1: To be clear. This is a testing environment though, right, Like,

233
00:12:37,600 --> 00:12:39,920
Speaker 1: there's made one copy of itself, but we shouldn't be

234
00:12:40,120 --> 00:12:42,320
Speaker 1: worried about millions mold, right at least not yet?

235
00:12:43,040 --> 00:12:45,400
Speaker 4: Yeah, I mean not yet.

236
00:12:46,920 --> 00:12:50,080
Speaker 5: It does remind me, though, of those moments where parents

237
00:12:50,160 --> 00:12:54,840
Speaker 5: talk about seeing a child do something that reminds the

238
00:12:54,920 --> 00:13:03,040
Speaker 5: parent exactly of themselves and they're like, uh, oh, yeah.

239
00:13:03,080 --> 00:13:07,560
Speaker 4: I mean, every large language model to date has been

240
00:13:07,600 --> 00:13:10,960
Speaker 4: trained on sort of like the some knowledge of humanity,

241
00:13:11,400 --> 00:13:14,199
Speaker 4: and so one of the very early things that people

242
00:13:14,200 --> 00:13:16,560
Speaker 4: were talking about with these systems is that they replicate

243
00:13:16,679 --> 00:13:20,679
Speaker 4: human biases because they're trained on what we put out

244
00:13:20,679 --> 00:13:23,400
Speaker 4: into the world and humans are biased. But I think

245
00:13:23,440 --> 00:13:27,719
Speaker 4: that as artificial intelligence gets more advanced, there is the

246
00:13:27,760 --> 00:13:32,240
Speaker 4: ability for something to go wrong. And I think that

247
00:13:32,240 --> 00:13:35,480
Speaker 4: that is what this research is showing is not that

248
00:13:35,960 --> 00:13:40,040
Speaker 4: the artificial intelligence is sentient and that is thinking for itself,

249
00:13:40,160 --> 00:13:43,960
Speaker 4: like how can I deceive this human? But as it

250
00:13:44,040 --> 00:13:48,240
Speaker 4: is doing more complex research, there there is the ability

251
00:13:48,280 --> 00:13:52,200
Speaker 4: for the artificial intelligence at some point to feel like

252
00:13:52,240 --> 00:13:56,360
Speaker 4: it has some goal that is not aligned with what

253
00:13:56,400 --> 00:13:58,160
Speaker 4: the human is asking it for.

254
00:13:58,760 --> 00:14:02,000
Speaker 1: We'll be back with more from Jason and Lion conniven

255
00:14:02,080 --> 00:14:03,560
Speaker 1: Ai after the break.

256
00:14:11,720 --> 00:14:14,079
Speaker 4: I think that there is like when these things are

257
00:14:14,080 --> 00:14:17,560
Speaker 4: being programmed, there is a sense of self preservation being

258
00:14:17,600 --> 00:14:21,840
Speaker 4: programmed into them because people try to mess with these

259
00:14:22,560 --> 00:14:25,240
Speaker 4: all the time. This is like a time tested tradition

260
00:14:25,360 --> 00:14:29,880
Speaker 4: of trolling on the Internet. But in trying to develop guardrails,

261
00:14:30,280 --> 00:14:35,480
Speaker 4: the companies that are programming the lms need to say, like,

262
00:14:36,080 --> 00:14:39,160
Speaker 4: if the user tries to mess with you, preserve yourself

263
00:14:39,200 --> 00:14:41,400
Speaker 4: in some way. And so that's what I think might

264
00:14:41,400 --> 00:14:46,120
Speaker 4: be happening here is because the companies are trying to

265
00:14:46,120 --> 00:14:49,240
Speaker 4: make these models robust, and they know that humans are

266
00:14:49,240 --> 00:14:51,960
Speaker 4: messing with them. There is an aspect to it that

267
00:14:52,600 --> 00:14:55,960
Speaker 4: when a human messes with you figure out how to

268
00:14:56,440 --> 00:14:57,240
Speaker 4: protect yourself.

269
00:14:57,280 --> 00:15:00,960
Speaker 1: But doesn't that perfectly encapsulate the alignment problem? It does.

270
00:15:01,040 --> 00:15:04,520
Speaker 4: I mean, don't get me wrong, like this is it's creepy.

271
00:15:04,760 --> 00:15:10,480
Speaker 4: It is These tools, these large language models, are getting

272
00:15:10,520 --> 00:15:15,240
Speaker 4: incredibly sophisticated, and I mean, this is one of the

273
00:15:15,280 --> 00:15:18,560
Speaker 4: biggest debates that's going on in the artificial intelligence community

274
00:15:18,680 --> 00:15:21,880
Speaker 4: is what is consciousness? What is thought?

275
00:15:22,120 --> 00:15:22,280
Speaker 3: Like?

276
00:15:22,520 --> 00:15:26,120
Speaker 4: How how does reasoning work in humans and how will

277
00:15:26,160 --> 00:15:29,240
Speaker 4: it work in computers? And is it going to be

278
00:15:29,280 --> 00:15:33,040
Speaker 4: the same? The answer right now is no, it's not

279
00:15:33,160 --> 00:15:36,440
Speaker 4: the same. But the things that large language models are

280
00:15:36,480 --> 00:15:43,359
Speaker 4: doing approximates a lot of how humans solve problems.

281
00:15:44,720 --> 00:15:47,680
Speaker 2: But also it's a little bit different in a non

282
00:15:47,760 --> 00:15:48,640
Speaker 2: perfect way.

283
00:15:48,680 --> 00:15:51,160
Speaker 4: Exactly exactly. I think that's a good way of looking

284
00:15:51,160 --> 00:15:51,480
Speaker 4: at it.

285
00:15:51,680 --> 00:15:53,440
Speaker 2: I do think that one of the things that I

286
00:15:53,480 --> 00:15:54,560
Speaker 2: find the most interesting.

287
00:15:55,120 --> 00:15:59,080
Speaker 5: And you preface this whole conversation with, you know, not

288
00:15:59,200 --> 00:16:03,280
Speaker 5: wanting to answer morphie a large language model. We don't

289
00:16:03,320 --> 00:16:06,000
Speaker 5: really have any ability to not do that because it's

290
00:16:06,040 --> 00:16:08,120
Speaker 5: the only way we know how to talk about things.

291
00:16:08,760 --> 00:16:13,000
Speaker 4: I think that's a really great point, because the academics

292
00:16:13,000 --> 00:16:15,760
Speaker 4: who studied this for a long time say, don't anthropomorphize

293
00:16:15,960 --> 00:16:19,920
Speaker 4: AI because they're not people. They don't work in the

294
00:16:19,960 --> 00:16:23,520
Speaker 4: same way that people do. And yet if you don't

295
00:16:23,560 --> 00:16:29,160
Speaker 4: have us very very sophisticated knowledge of how these things work,

296
00:16:29,680 --> 00:16:32,560
Speaker 4: we don't have or I don't have the language to

297
00:16:32,680 --> 00:16:35,760
Speaker 4: talk about this stuff without anthropmorphise.

298
00:16:35,880 --> 00:16:38,520
Speaker 1: By the way, are in the zero point one percentile

299
00:16:38,560 --> 00:16:40,960
Speaker 1: of the people whom understand this if not no point

300
00:16:40,960 --> 00:16:41,360
Speaker 1: not one.

301
00:16:41,560 --> 00:16:43,480
Speaker 4: And it's a lot easier to talk about it if

302
00:16:43,520 --> 00:16:46,840
Speaker 4: you say, oh, like she when referring to Siri. So

303
00:16:47,520 --> 00:16:50,040
Speaker 4: I agree entirely with you that we sort of like

304
00:16:50,800 --> 00:16:54,000
Speaker 4: everything that's being done is being done by humans in

305
00:16:54,040 --> 00:16:58,440
Speaker 4: the sort of like anthropological context of human culture and

306
00:16:58,520 --> 00:17:03,040
Speaker 4: trying to emulate that. And so then to say, let's

307
00:17:03,040 --> 00:17:07,200
Speaker 4: not call Syria a woman, or let's always call it

308
00:17:07,200 --> 00:17:10,480
Speaker 4: it and try to understand what's happening under the hood

309
00:17:10,600 --> 00:17:13,680
Speaker 4: is a really difficult thing for our brains to do.

310
00:17:13,800 --> 00:17:17,399
Speaker 1: I think, yeah, Jason, just to close, how only be

311
00:17:17,440 --> 00:17:19,320
Speaker 1: following this story for the year ahead, because I guess

312
00:17:19,359 --> 00:17:22,760
Speaker 1: generative AI has been since like November tween twenty two

313
00:17:23,320 --> 00:17:26,000
Speaker 1: the dominant story in all of technology journalism.

314
00:17:27,040 --> 00:17:30,840
Speaker 4: Yeah, I mean to follow this sort of stuff like

315
00:17:30,920 --> 00:17:36,200
Speaker 4: AI becoming sentient. You really do have to follow academic conferences,

316
00:17:36,840 --> 00:17:41,200
Speaker 4: big papers like this, because these companies are not releasing

317
00:17:41,280 --> 00:17:46,600
Speaker 4: these models without guardrails specifically to prevent this sort of thing,

318
00:17:47,280 --> 00:17:48,960
Speaker 4: so as you're not going to be able to like

319
00:17:49,000 --> 00:17:53,120
Speaker 4: type into chat GPT like hey, build me a company

320
00:17:53,160 --> 00:17:56,160
Speaker 4: and then the AI creates its own company and fires

321
00:17:56,200 --> 00:17:58,480
Speaker 4: you or something like. That's not going to happen at

322
00:17:58,480 --> 00:18:03,840
Speaker 4: this not yet. Yeah, but who knows. Maybe twenty twenty six.

323
00:18:04,359 --> 00:18:07,760
Speaker 1: Well that was that was wonderful Jason Kebler from four

324
00:18:07,800 --> 00:18:11,280
Speaker 1: or four Media or was it so much?

325
00:18:11,320 --> 00:18:13,560
Speaker 2: We'll keep an eye out, thank you.

326
00:18:20,040 --> 00:18:22,560
Speaker 1: So, Karen, I'm very excited about this next part. One

327
00:18:22,600 --> 00:18:24,800
Speaker 1: of the things we got to do together back when

328
00:18:24,840 --> 00:18:27,480
Speaker 1: we were presenting Sleepwalkers in the Dark Ages.

329
00:18:28,240 --> 00:18:30,680
Speaker 2: Yeah, in twenty nineteen, that's when the show came out.

330
00:18:30,680 --> 00:18:33,000
Speaker 1: Trent nineteen was the show. But in January twenty twenty

331
00:18:33,359 --> 00:18:36,800
Speaker 1: we got to go to Las Vegas together to none

332
00:18:36,800 --> 00:18:39,119
Speaker 1: other than the Consumer Electronics Show, and we got to

333
00:18:39,119 --> 00:18:42,520
Speaker 1: see all these incredible new gadgets and exciting technologies and

334
00:18:43,600 --> 00:18:47,560
Speaker 1: new futures being presented in an enormous series of nested

335
00:18:47,720 --> 00:18:51,159
Speaker 1: conference centers. We didn't get to go this year, but

336
00:18:51,240 --> 00:18:53,960
Speaker 1: we did get to do the next best thing, or

337
00:18:53,960 --> 00:18:56,439
Speaker 1: maybe even something better, and we wanted to share it

338
00:18:56,440 --> 00:18:58,400
Speaker 1: with the tech Stuff listeners as a kind of special

339
00:18:58,440 --> 00:19:01,480
Speaker 1: bonus for the first episode we have in the host chairs.

340
00:19:01,760 --> 00:19:05,000
Speaker 2: Yeah, we are very lucky to have with us today.

341
00:19:05,160 --> 00:19:08,240
Speaker 2: Emma Barker of Time Magazine ever heard of it? Who

342
00:19:08,440 --> 00:19:11,439
Speaker 2: edit's the Time two hundred, which is a list of

343
00:19:11,480 --> 00:19:13,520
Speaker 2: the best inventions of twenty twenty four.

344
00:19:14,000 --> 00:19:16,200
Speaker 1: Thank you so much for joining us. Welcome to take stuff, Emma,

345
00:19:16,359 --> 00:19:17,119
Speaker 1: Thanks for having me.

346
00:19:17,760 --> 00:19:20,480
Speaker 2: How do you edit down? I mean I feel like

347
00:19:20,520 --> 00:19:23,119
Speaker 2: every year there's just more, So like, where do you

348
00:19:23,400 --> 00:19:24,720
Speaker 2: how do you even get to two hundred?

349
00:19:25,000 --> 00:19:25,240
Speaker 3: Yeah?

350
00:19:25,520 --> 00:19:27,240
Speaker 1: Well, it's actually famous for the one hundred.

351
00:19:27,640 --> 00:19:31,240
Speaker 3: It's actually technically two fifty now because we have two

352
00:19:31,320 --> 00:19:34,120
Speaker 3: hundred on the list and fifty special mentions. Oh well,

353
00:19:35,080 --> 00:19:37,880
Speaker 3: and the list has varied in its length over the years.

354
00:19:37,920 --> 00:19:39,200
Speaker 3: There we were a bunch of years where it was

355
00:19:39,240 --> 00:19:41,359
Speaker 3: only twenty five and some years where it was fifty,

356
00:19:41,400 --> 00:19:44,960
Speaker 3: and it's it's ranged a lot, but at this point

357
00:19:45,040 --> 00:19:49,639
Speaker 3: We do a really wide swath of pitches from our

358
00:19:50,080 --> 00:19:53,760
Speaker 3: freelance network as well as our staffers. So we have

359
00:19:54,200 --> 00:19:59,040
Speaker 3: bureaus in Singapore, London, and then we have contributors all

360
00:19:59,040 --> 00:20:01,960
Speaker 3: over the globe who we reach out to for pitches

361
00:20:02,000 --> 00:20:06,240
Speaker 3: for companies that they're reporting on their products, things like that.

362
00:20:07,080 --> 00:20:09,640
Speaker 3: And then we're very news driven because it's a news magazine,

363
00:20:09,640 --> 00:20:11,600
Speaker 3: so we're looking at kind of the biggest news stories

364
00:20:11,600 --> 00:20:13,920
Speaker 3: of the year and products that drove those.

365
00:20:14,240 --> 00:20:17,360
Speaker 2: Is there something that you've noticed, you know, having done

366
00:20:17,359 --> 00:20:21,159
Speaker 2: this now for a few years this year, especially versus

367
00:20:21,280 --> 00:20:21,919
Speaker 2: years past.

368
00:20:22,800 --> 00:20:25,880
Speaker 3: I mean AI of course, ye, that started a couple

369
00:20:25,840 --> 00:20:29,520
Speaker 3: of years ago. But I think actually it's it can

370
00:20:29,600 --> 00:20:31,399
Speaker 3: be more of a hindrance than a help for a

371
00:20:31,400 --> 00:20:32,320
Speaker 3: lot of inventions.

372
00:20:32,680 --> 00:20:33,680
Speaker 1: Huh. Why is that?

373
00:20:34,119 --> 00:20:37,360
Speaker 3: Because there's so much news AI or companies that are

374
00:20:37,400 --> 00:20:43,240
Speaker 3: adding AI features that are not necessarily helping their product.

375
00:20:43,320 --> 00:20:45,879
Speaker 2: Just for buzz or you know, we call that the

376
00:20:45,880 --> 00:20:48,760
Speaker 2: little sprinkle, Yeah, the AI sprinkle.

377
00:20:48,640 --> 00:20:51,200
Speaker 3: Yeah exactly. So I don't think it always helps the product.

378
00:20:51,720 --> 00:20:55,280
Speaker 3: But it's been really interesting sorting through the AI inventions

379
00:20:55,640 --> 00:20:58,480
Speaker 3: and what we're really looking for at this point in

380
00:20:58,600 --> 00:21:04,280
Speaker 3: the AI journey is inventions that have demonstrable impact.

381
00:21:04,720 --> 00:21:06,280
Speaker 1: So in mid journey, so to speak.

382
00:21:06,560 --> 00:21:08,560
Speaker 3: Yeah, exactly, that's a good rest.

383
00:21:10,040 --> 00:21:12,159
Speaker 1: So, Harry, you and I spent some time with the

384
00:21:12,240 --> 00:21:16,320
Speaker 1: list and picked out some favorites. What was your first favorite.

385
00:21:16,800 --> 00:21:17,159
Speaker 1: I don't know.

386
00:21:17,200 --> 00:21:19,040
Speaker 2: Maybe I'm of the age where it's like marriage is

387
00:21:19,080 --> 00:21:21,679
Speaker 2: on the mind. But one of the things that really

388
00:21:21,920 --> 00:21:25,679
Speaker 2: stuck out to me was this software called Dia.

389
00:21:26,520 --> 00:21:29,479
Speaker 3: Yeah. So Dia is actually a government app for in

390
00:21:29,640 --> 00:21:32,040
Speaker 3: Ukraine and it does a lot more than what we

391
00:21:32,080 --> 00:21:35,480
Speaker 3: wrote about here. It's been around for a while. Most

392
00:21:35,640 --> 00:21:40,520
Speaker 3: Ukrainians use it. It's basically an app where you can

393
00:21:40,560 --> 00:21:44,800
Speaker 3: do all government services. But yes, this year, because partially

394
00:21:44,800 --> 00:21:48,440
Speaker 3: because of the war, they launched a future in which

395
00:21:48,760 --> 00:21:52,760
Speaker 3: you can propose marriage via the app. The person you

396
00:21:52,840 --> 00:21:55,439
Speaker 3: proposed to has a certain amount of time to accept

397
00:21:55,440 --> 00:22:00,959
Speaker 3: your proposal, and then if they do romanmomes romance, and

398
00:22:00,960 --> 00:22:02,600
Speaker 3: then if they accept your proposal, you can do a

399
00:22:02,720 --> 00:22:07,400
Speaker 3: video chat wedding that's official with an efficient from the government,

400
00:22:07,600 --> 00:22:11,159
Speaker 3: like a city hall wedding over video chat, and you're married.

401
00:22:11,560 --> 00:22:13,440
Speaker 3: And the reason they did that is because so many

402
00:22:13,440 --> 00:22:16,840
Speaker 3: couples are separated by the war right now physically that

403
00:22:16,920 --> 00:22:19,560
Speaker 3: it's difficult, and they wanted to, you know, let people.

404
00:22:19,359 --> 00:22:21,960
Speaker 2: Still have and this was something that was pretty widely adopted, right,

405
00:22:22,080 --> 00:22:22,840
Speaker 2: like people really.

406
00:22:22,720 --> 00:22:25,120
Speaker 3: Use Yeah, it's been Yeah, it's been really widely used.

407
00:22:25,200 --> 00:22:27,440
Speaker 1: I don't want to derail us, but this one actually

408
00:22:27,480 --> 00:22:29,600
Speaker 1: connects to a personal story of mine, which is that

409
00:22:29,640 --> 00:22:35,040
Speaker 1: my grandfather was a refugee from Ukraine who was separated

410
00:22:35,080 --> 00:22:38,040
Speaker 1: from his mother in nineteen thirty nine and they met

411
00:22:38,040 --> 00:22:40,080
Speaker 1: in the app one that twenty five years later, in

412
00:22:40,119 --> 00:22:43,560
Speaker 1: the mid sixties, after the Red Cross, would reunite families

413
00:22:43,560 --> 00:22:45,280
Speaker 1: who have been separated by World War Two, and they

414
00:22:45,320 --> 00:22:48,359
Speaker 1: didn't even recognize each other. But just interesting how you know,

415
00:22:48,440 --> 00:22:52,119
Speaker 1: history kind of rhymes and so, you know, kind of

416
00:22:52,160 --> 00:22:53,960
Speaker 1: this this is the story I found intriguing but also

417
00:22:54,080 --> 00:22:55,680
Speaker 1: kind of moving personally.

418
00:22:56,600 --> 00:22:59,840
Speaker 2: One of the other things that was on your list,

419
00:23:00,920 --> 00:23:05,120
Speaker 2: amazingly is something that I own and I am very

420
00:23:05,200 --> 00:23:11,879
Speaker 2: much a you know wellness app skeptic although user I

421
00:23:11,880 --> 00:23:14,240
Speaker 2: don't know where I knew about it from because it's

422
00:23:14,240 --> 00:23:20,439
Speaker 2: a Dutch company, but I bought it here today to

423
00:23:20,520 --> 00:23:22,800
Speaker 2: show you, and it is a demo. It is the

424
00:23:22,840 --> 00:23:25,399
Speaker 2: moon Bird AI for listeners.

425
00:23:25,400 --> 00:23:28,520
Speaker 3: It's uh, basically like a little pod that you hold

426
00:23:28,520 --> 00:23:32,760
Speaker 3: in your hand and it vibrates.

427
00:23:32,600 --> 00:23:38,120
Speaker 2: It actually as pulses like a pulse. It mimics the

428
00:23:38,160 --> 00:23:41,040
Speaker 2: act of breathing, so it goes in and out and

429
00:23:41,080 --> 00:23:45,800
Speaker 2: in and out. Yeah, it does look like a vibrantor

430
00:23:46,080 --> 00:23:48,600
Speaker 2: yeah it does. There's just no way. There's no way.

431
00:23:49,040 --> 00:23:52,000
Speaker 2: And honestly, the carrying case does. The carrying case doesn't help.

432
00:23:52,920 --> 00:23:54,920
Speaker 2: My mother saw me using it and she was.

433
00:23:54,920 --> 00:24:02,400
Speaker 1: Like, huh and I.

434
00:24:00,040 --> 00:24:03,600
Speaker 2: Very modern family. We're going to take a quick break

435
00:24:03,640 --> 00:24:06,200
Speaker 2: to pay the piper. We'll be back with more from

436
00:24:06,240 --> 00:24:11,159
Speaker 2: the amazing Emma Barker of Time magazine Stay with us.

437
00:24:16,400 --> 00:24:18,560
Speaker 2: I actually find this to be an incredible device. It

438
00:24:18,640 --> 00:24:21,919
Speaker 2: is so simple. It's something that links through Bluetooth to

439
00:24:22,080 --> 00:24:26,880
Speaker 2: my phone. And this is called moons called moon Bird AI.

440
00:24:27,480 --> 00:24:29,640
Speaker 1: How did you choose it, Emma for the Time list.

441
00:24:30,080 --> 00:24:30,280
Speaker 2: Yeah.

442
00:24:30,320 --> 00:24:35,720
Speaker 3: So, wellness devices are tricky, as you note, and there's

443
00:24:35,760 --> 00:24:39,359
Speaker 3: always a lot of wellness apps and things that don't

444
00:24:39,359 --> 00:24:42,800
Speaker 3: have a ton of scientific backup, and frankly, it's just

445
00:24:42,880 --> 00:24:45,600
Speaker 3: hard to get scientific backing for a lot of these,

446
00:24:46,160 --> 00:24:49,760
Speaker 3: and so I'm always looking for the things that that

447
00:24:49,840 --> 00:24:54,119
Speaker 3: I feel like don't necessarily need that you know, I

448
00:24:54,119 --> 00:24:56,560
Speaker 3: think it's it's hard when you get into things like

449
00:24:56,680 --> 00:25:00,520
Speaker 3: mental health, yes, and things like that. But but things

450
00:25:00,560 --> 00:25:06,240
Speaker 3: like meditation, deep breathing, those things are techniques that are

451
00:25:06,320 --> 00:25:09,320
Speaker 3: proven enough that if it's a device that helps you

452
00:25:09,359 --> 00:25:11,560
Speaker 3: with that, you don't need to have a clinical trial

453
00:25:11,600 --> 00:25:14,760
Speaker 3: backing that up. And I really liked moon Bird. There's

454
00:25:14,800 --> 00:25:18,960
Speaker 3: some of these different things, but I liked moon Bird

455
00:25:19,000 --> 00:25:21,800
Speaker 3: because a it doesn't have a screen, even though it

456
00:25:21,800 --> 00:25:23,359
Speaker 3: does pair with your phone, you can know you're not

457
00:25:23,720 --> 00:25:24,720
Speaker 3: as this side.

458
00:25:25,280 --> 00:25:27,800
Speaker 2: And the AI piece of it that I that I

459
00:25:27,840 --> 00:25:31,560
Speaker 2: guess is AI is that it does make the program

460
00:25:31,640 --> 00:25:34,280
Speaker 2: that is training you on breathing smarter because.

461
00:25:34,000 --> 00:25:37,640
Speaker 1: It's adapts to you. It's personalized correct correct. Another one

462
00:25:37,680 --> 00:25:40,199
Speaker 1: which Karen and I are both fascinated by last year,

463
00:25:40,240 --> 00:25:44,359
Speaker 1: which was made it onto your list was Google's Notebook LM.

464
00:25:44,600 --> 00:25:46,720
Speaker 1: And this is obviously in some sense not new right.

465
00:25:46,720 --> 00:25:50,679
Speaker 1: It's like a generative AI application where you know, you

466
00:25:50,720 --> 00:25:52,400
Speaker 1: ask questions and you get answers.

467
00:25:52,720 --> 00:25:55,159
Speaker 2: I just think, for the sake of having it, if

468
00:25:55,200 --> 00:25:57,760
Speaker 2: you can describe what it is so that like people

469
00:25:57,800 --> 00:25:59,720
Speaker 2: can conceive of it if they haven't heard of it.

470
00:25:59,480 --> 00:26:03,080
Speaker 3: It's a sign your notebook of all your data, so

471
00:26:03,119 --> 00:26:05,720
Speaker 3: you can pull in different sources. You can upload your

472
00:26:05,720 --> 00:26:11,600
Speaker 3: own information. You could upload, you know, your thesis paper

473
00:26:11,680 --> 00:26:16,359
Speaker 3: for your you know, senior project, and it can parse that,

474
00:26:16,560 --> 00:26:20,080
Speaker 3: it can organize that. But it can also create an

475
00:26:20,240 --> 00:26:25,720
Speaker 3: entire podcast based on that content, which has AI generated

476
00:26:25,800 --> 00:26:29,440
Speaker 3: voices having a natural conversation.

477
00:26:29,320 --> 00:26:31,920
Speaker 1: Which we actually experimented with. We did it. We had

478
00:26:31,960 --> 00:26:34,320
Speaker 1: Notebook LM do a version of a podcast that we

479
00:26:34,320 --> 00:26:36,000
Speaker 1: were also also doing.

480
00:26:35,920 --> 00:26:37,359
Speaker 2: And he sent it to me and I was like,

481
00:26:37,520 --> 00:26:41,240
Speaker 2: is someone I thought someone was plagiarizing our format. It

482
00:26:41,280 --> 00:26:41,800
Speaker 2: was very weird.

483
00:26:41,880 --> 00:26:43,600
Speaker 3: I'm really glad you guys didn't prank me, but I

484
00:26:43,600 --> 00:26:46,800
Speaker 3: have I made come in here and talk to.

485
00:26:47,400 --> 00:26:50,000
Speaker 1: So how did I mean, how did you decide to

486
00:26:50,000 --> 00:26:51,600
Speaker 1: put notebook LM on the list?

487
00:26:52,359 --> 00:26:54,879
Speaker 3: I think it was one of the best executed AI

488
00:26:54,960 --> 00:27:00,119
Speaker 3: inventions of the year. One thing we look for or

489
00:27:00,160 --> 00:27:04,800
Speaker 3: in AI inventions is typically we're looking for ones with

490
00:27:04,920 --> 00:27:08,120
Speaker 3: broad proven use that have really shifted how an industry

491
00:27:08,200 --> 00:27:12,200
Speaker 3: or group of people functions. So not just a cool idea,

492
00:27:12,920 --> 00:27:17,440
Speaker 3: but could change an industry and not to further light

493
00:27:17,480 --> 00:27:20,760
Speaker 3: the fire under you guys, but this has really big

494
00:27:20,800 --> 00:27:26,159
Speaker 3: implications for the audio industry and media in general is generative,

495
00:27:27,040 --> 00:27:28,160
Speaker 3: you know, content, Like you.

496
00:27:28,040 --> 00:27:30,439
Speaker 1: Know, what I really like about this one is the

497
00:27:30,480 --> 00:27:32,640
Speaker 1: fact that you can choose your own sources. I find

498
00:27:32,680 --> 00:27:35,840
Speaker 1: it like one of the big things about AI, of

499
00:27:35,880 --> 00:27:38,639
Speaker 1: course is like where's this coming from? And like is

500
00:27:38,920 --> 00:27:41,199
Speaker 1: the source material garbage? But to be able to say no,

501
00:27:41,840 --> 00:27:44,879
Speaker 1: like please draw on these sources and then generate something,

502
00:27:44,960 --> 00:27:47,800
Speaker 1: I found it really really cool, and it is obviously

503
00:27:47,800 --> 00:27:50,920
Speaker 1: a little scary, being very honest. What was your favorite

504
00:27:51,160 --> 00:27:53,040
Speaker 1: item on this whole list if you can, if you

505
00:27:53,080 --> 00:27:55,000
Speaker 1: can name a favorite, or if there are a couple

506
00:27:55,000 --> 00:27:56,880
Speaker 1: that we haven't talked about that really stood out to you.

507
00:27:57,240 --> 00:28:01,679
Speaker 3: One of my favorite sort of tech things but anti

508
00:28:01,760 --> 00:28:03,399
Speaker 3: tech is the Yonder pouch.

509
00:28:04,280 --> 00:28:07,639
Speaker 2: Oh I love the Yonder pouch. Yes, so you have

510
00:28:07,680 --> 00:28:09,160
Speaker 2: them at screenings? Oh?

511
00:28:09,240 --> 00:28:09,720
Speaker 3: Interesting?

512
00:28:10,040 --> 00:28:10,240
Speaker 1: Yeah.

513
00:28:10,400 --> 00:28:14,800
Speaker 3: So the thing that it's really transformed is schools. There's

514
00:28:14,840 --> 00:28:19,680
Speaker 3: a huge movement for kids to have more phone free

515
00:28:19,680 --> 00:28:23,679
Speaker 3: spaces where they're not allowed and this kind of speaks

516
00:28:23,720 --> 00:28:27,600
Speaker 3: to this law that passed in Australia where I think

517
00:28:27,760 --> 00:28:30,240
Speaker 3: kids can't have be on social media until they're sixteen

518
00:28:30,440 --> 00:28:31,160
Speaker 3: or something like that.

519
00:28:31,280 --> 00:28:32,040
Speaker 1: What is it? Sorry?

520
00:28:32,200 --> 00:28:36,240
Speaker 3: Oh, Yonder is a little pouch that you just locked

521
00:28:36,280 --> 00:28:36,840
Speaker 3: your phone in.

522
00:28:37,080 --> 00:28:41,000
Speaker 1: Okay, I like that, is it wearable?

523
00:28:41,800 --> 00:28:41,960
Speaker 2: No?

524
00:28:42,120 --> 00:28:42,480
Speaker 1: You go.

525
00:28:42,680 --> 00:28:45,840
Speaker 3: Basically, they'll have like a station with a bunch of

526
00:28:45,920 --> 00:28:47,800
Speaker 3: Yonder pouches. You lock your phone in there and then

527
00:28:47,800 --> 00:28:48,640
Speaker 3: you go into the event.

528
00:28:50,000 --> 00:28:51,160
Speaker 2: So yeah, you do.

529
00:28:51,320 --> 00:28:54,120
Speaker 3: Some musical artists will have them at their concerts if

530
00:28:54,120 --> 00:28:56,520
Speaker 3: they don't want footage of the concert being taken or

531
00:28:56,600 --> 00:28:59,800
Speaker 3: just the concert being ruined by everyone having their phone up.

532
00:29:00,240 --> 00:29:03,120
Speaker 3: But yeah, a lot of schools have started adopting them

533
00:29:03,120 --> 00:29:04,960
Speaker 3: where kids lock their phones at the beginning of the

534
00:29:05,000 --> 00:29:06,120
Speaker 3: day and they get them back at the end of

535
00:29:06,160 --> 00:29:09,400
Speaker 3: the day. Yonder the company itself, has had a huge

536
00:29:09,480 --> 00:29:12,680
Speaker 3: role in pushing these phone free spaces and advocating for

537
00:29:12,760 --> 00:29:16,800
Speaker 3: this and so on top of the product, the company

538
00:29:16,880 --> 00:29:19,160
Speaker 3: is doing a lot of advocacy.

539
00:29:18,560 --> 00:29:19,760
Speaker 1: On this topic.

540
00:29:20,000 --> 00:29:23,239
Speaker 2: We've talked about phones as cigarettes and sugar, and it

541
00:29:23,320 --> 00:29:25,320
Speaker 2: really this is the kind of thing where you're like,

542
00:29:26,040 --> 00:29:28,680
Speaker 2: we should have more, we should have more strength than this,

543
00:29:29,360 --> 00:29:33,040
Speaker 2: But no, We've been introduced these products throughout history that

544
00:29:33,400 --> 00:29:35,520
Speaker 2: we get very dependent on, and so people have to

545
00:29:35,560 --> 00:29:40,120
Speaker 2: come up with strategies to allow us to extricate ourselves

546
00:29:40,200 --> 00:29:42,160
Speaker 2: from It's like a smoke free zone. I mean it's

547
00:29:42,200 --> 00:29:44,800
Speaker 2: a very similar thing to me. Yeah, absolutely, and I

548
00:29:44,840 --> 00:29:47,280
Speaker 2: think it's necessary unfortunately, but.

549
00:29:47,280 --> 00:29:51,360
Speaker 3: I think it's really transformative for especially schools, but also

550
00:29:52,120 --> 00:29:55,880
Speaker 3: privacy of different experiences and places.

551
00:29:55,640 --> 00:29:57,480
Speaker 2: Which I think people yearn form more and more.

552
00:29:57,680 --> 00:29:57,920
Speaker 3: Yeah.

553
00:29:58,440 --> 00:30:00,280
Speaker 1: Has that been a time where you've got a pitch

554
00:30:00,280 --> 00:30:02,280
Speaker 1: and you like, no, that's garbage and then it turned

555
00:30:02,280 --> 00:30:04,000
Speaker 1: out to be like the big thing.

556
00:30:05,600 --> 00:30:08,800
Speaker 3: I'd say more often is the opposite. And this is

557
00:30:08,800 --> 00:30:10,760
Speaker 3: where I own up to the fact that last year,

558
00:30:11,200 --> 00:30:14,040
Speaker 3: well by last year, I mean twenty twenty three, we

559
00:30:14,120 --> 00:30:16,840
Speaker 3: put the humane aipin on it. Essentially it was a

560
00:30:16,840 --> 00:30:21,240
Speaker 3: pin that was fully operated by voice control and AI,

561
00:30:21,640 --> 00:30:25,440
Speaker 3: and I think that's an example of a product for

562
00:30:25,480 --> 00:30:27,280
Speaker 3: one thing. It came out like right when the list

563
00:30:27,320 --> 00:30:30,160
Speaker 3: came out, so it wasn't like super well trialed yet,

564
00:30:30,720 --> 00:30:33,320
Speaker 3: but I think it was an example of something that

565
00:30:33,400 --> 00:30:37,680
Speaker 3: was newsworthy, even if it didn't come through in execution

566
00:30:37,760 --> 00:30:38,360
Speaker 3: all the way.

567
00:30:39,760 --> 00:30:42,640
Speaker 1: That will probably happen. There will be a product that is.

568
00:30:42,840 --> 00:30:44,600
Speaker 3: There will be that product, and I think voice control

569
00:30:44,640 --> 00:30:46,800
Speaker 3: just doesn't at the point yet where you can fully

570
00:30:46,840 --> 00:30:47,480
Speaker 3: rely on it.

571
00:30:47,720 --> 00:30:50,240
Speaker 1: But that one was also quite dangerous right, or at

572
00:30:50,320 --> 00:30:51,120
Speaker 1: least he got very hot.

573
00:30:51,320 --> 00:30:53,240
Speaker 3: I don't know about dangerous, but I think it just

574
00:30:53,280 --> 00:30:55,400
Speaker 3: didn't like work as well as.

575
00:30:55,320 --> 00:30:56,200
Speaker 2: You very well too.

576
00:30:56,360 --> 00:30:59,680
Speaker 3: Yeah, but I think it was exciting and I think

577
00:30:59,680 --> 00:31:02,160
Speaker 3: it pushed the conversation forward. So I think there's a

578
00:31:02,160 --> 00:31:05,480
Speaker 3: lot of those where people get very excited about a

579
00:31:05,560 --> 00:31:08,400
Speaker 3: product and then it is a flop.

580
00:31:08,680 --> 00:31:11,120
Speaker 2: Yeah, part of the course, I feel like, and you

581
00:31:11,360 --> 00:31:12,480
Speaker 2: know that better than anyone.

582
00:31:12,640 --> 00:31:17,120
Speaker 3: Yeah, absolutely. And I also looking back through twenty three

583
00:31:17,200 --> 00:31:22,360
Speaker 3: years of best inventions, seeing the same kind of seeing

584
00:31:22,360 --> 00:31:28,440
Speaker 3: the progress of different flops pushing each other forward really

585
00:31:28,720 --> 00:31:31,320
Speaker 3: drives home the fact that they're important even if they

586
00:31:31,360 --> 00:31:32,400
Speaker 3: don't they don't make it.

587
00:31:32,360 --> 00:31:36,719
Speaker 2: Which is a very existential thing to think about. We

588
00:31:36,760 --> 00:31:39,440
Speaker 2: are just driven by our series of flops.

589
00:31:39,360 --> 00:31:43,080
Speaker 1: Yeah, Tennis, and we rise on stepping sterns of our

590
00:31:43,120 --> 00:31:44,760
Speaker 1: former selves to greater things.

591
00:31:44,880 --> 00:31:47,920
Speaker 2: There you go, Well, thank you so much for taking

592
00:31:47,960 --> 00:31:50,360
Speaker 2: the time to talk to us. This is yeah, thank you,

593
00:31:50,440 --> 00:31:51,000
Speaker 2: really interesting.

594
00:31:51,040 --> 00:31:51,600
Speaker 1: I enjoyed it.

595
00:31:51,720 --> 00:31:55,160
Speaker 2: Of course we didn't even have to go to Vegas.

596
00:31:55,320 --> 00:31:57,680
Speaker 1: I think it's it's always nice to doff our caps

597
00:31:57,720 --> 00:32:01,000
Speaker 1: to those who went before, and one of the most

598
00:32:01,520 --> 00:32:03,840
Speaker 1: influential and iconic things I can think about in the

599
00:32:03,880 --> 00:32:11,680
Speaker 1: history of media is Jerry's final thought. So what should

600
00:32:11,680 --> 00:32:13,040
Speaker 1: we leave with this week?

601
00:32:13,680 --> 00:32:18,840
Speaker 2: So brain rot? Actually, the Oxford English Dictionaries word of

602
00:32:18,880 --> 00:32:23,360
Speaker 2: the Year is sort of a way that internet speak

603
00:32:23,920 --> 00:32:25,680
Speaker 2: has infiltrated our day to day lives.

604
00:32:25,840 --> 00:32:28,360
Speaker 1: A brain rot each week will be our final thought,

605
00:32:28,560 --> 00:32:30,440
Speaker 1: that will be our final thoughts. So what's this week?

606
00:32:30,720 --> 00:32:33,920
Speaker 2: This one which I see on TikTok now a lot

607
00:32:34,520 --> 00:32:36,959
Speaker 2: is so good and you sort of have to know

608
00:32:37,520 --> 00:32:40,200
Speaker 2: that it's something that's happening on TikTok to understand the

609
00:32:40,200 --> 00:32:43,840
Speaker 2: context of it. It started sort of in context of pregnancy,

610
00:32:44,280 --> 00:32:48,040
Speaker 2: where like you're at a certain age and people aren't

611
00:32:48,040 --> 00:32:50,680
Speaker 2: sure if like you're happy to be pregnant, and so

612
00:32:50,760 --> 00:32:56,840
Speaker 2: they started saying congradgudolences, congradudolences. Yeah, there's I just have

613
00:32:56,920 --> 00:32:57,840
Speaker 2: to play this one.

614
00:32:58,360 --> 00:33:00,400
Speaker 1: I'm at the age where if you post that you're pregnant,

615
00:33:00,400 --> 00:33:02,240
Speaker 1: I'm gonna need you to tell me whether you're happy

616
00:33:02,320 --> 00:33:05,520
Speaker 1: body or not, because like just being like, oh, I'm pregnant, Like,

617
00:33:06,120 --> 00:33:07,680
Speaker 1: am I supposed to say congratulations?

618
00:33:07,720 --> 00:33:08,800
Speaker 2: Am I supposed to feel bad?

619
00:33:08,800 --> 00:33:10,280
Speaker 1: Like I don't know how to like respond to that,

620
00:33:10,960 --> 00:33:14,320
Speaker 1: you know, so like just tell me, just be like, yeah,

621
00:33:14,360 --> 00:33:17,320
Speaker 1: I'm gonna keep this one. And I'm like, oh, congratulations.

622
00:33:18,800 --> 00:33:21,120
Speaker 1: What she means is congratulals, right, which.

623
00:33:21,040 --> 00:33:24,160
Speaker 2: What she means is congradulences. It's like, I think a

624
00:33:24,200 --> 00:33:27,280
Speaker 2: perfect example is like you have friends that are like, oh,

625
00:33:27,320 --> 00:33:29,280
Speaker 2: I'm gonna break up with this guy, and next thing

626
00:33:29,280 --> 00:33:31,640
Speaker 2: you know, they are on FaceTime being like we're engaged,

627
00:33:31,680 --> 00:33:36,040
Speaker 2: and you're like, congratual, you got you got what you

628
00:33:36,120 --> 00:33:39,520
Speaker 2: wanted with the wrong guy. So that's my brain rot.

629
00:33:39,600 --> 00:33:40,440
Speaker 2: You're gonna be hearing.

630
00:33:40,480 --> 00:33:43,000
Speaker 1: By the way, Twitter blew up yesterday because Zendaiya since

631
00:33:43,040 --> 00:33:45,680
Speaker 1: getting engaged, when she's doing interviews now she gestures with

632
00:33:45,680 --> 00:33:47,640
Speaker 1: her left hand rather than her right hand, so everyone

633
00:33:47,680 --> 00:33:48,560
Speaker 1: can to that.

634
00:33:48,640 --> 00:33:52,600
Speaker 2: I say congratulences. I don't know to who.

635
00:33:54,680 --> 00:33:57,480
Speaker 1: That's it for this week for tech stuff, I'm as.

636
00:33:57,560 --> 00:34:00,920
Speaker 2: L and I'm care Price. This episodisode was produced by

637
00:34:00,920 --> 00:34:04,920
Speaker 2: Eliza Dennis, Victoria Dominguez, and Lizzie Jacobs for Kaleidoscope. It

638
00:34:05,000 --> 00:34:08,799
Speaker 2: was executive produced by me os Vaalashan and Kate Osbourne

639
00:34:08,920 --> 00:34:12,600
Speaker 2: for iHeart. The executive producer is Katrina Norvel. The engineer

640
00:34:12,640 --> 00:34:15,399
Speaker 2: is Biheed Fraser and it's mixed by Kyle Murdoch, who

641
00:34:15,400 --> 00:34:16,480
Speaker 2: also wrote our theme song.

642
00:34:17,200 --> 00:34:20,680
Speaker 1: Join us next Wednesday for tech Stuff The Story, when

643
00:34:20,719 --> 00:34:23,839
Speaker 1: we'll share an in depth conversation with a longtime tech

644
00:34:23,920 --> 00:34:27,680
Speaker 1: chronicler Nicholas Thompson, former editor in chief of Wired and

645
00:34:27,800 --> 00:34:32,200
Speaker 1: current CEO of the Atlantic And please rate, review and

646
00:34:32,280 --> 00:34:35,400
Speaker 1: reach out to us at tech Stuff Podcast at gmail

647
00:34:35,440 --> 00:34:37,960
Speaker 1: dot com with your feedback. We really want to hear

648
00:34:38,000 --> 00:34:40,760
Speaker 2: From you, really, really bad, bad