www-theatlantic-com-qqbuyc.txt (17907B)
1 [1]Skip to content 2 3 Site Navigation 4 5 • [2] 6 • 7 [5]Popular[6]Latest[7]Newsletters 8 9 Sections 10 11 □ [8]Politics 12 □ [9]Ideas 13 □ [10]Fiction 14 □ [11]Technology 15 □ [12]Science 16 □ [13]Photo 17 □ [14]Business 18 □ [15]Culture 19 □ [16]Planet 20 □ [17]Global 21 □ [18]Books 22 □ [19]Podcasts 23 □ [20]Health 24 □ [21]Education 25 □ [22]Projects 26 □ [23]Features 27 □ [24]Family 28 □ [25]Events 29 □ [26]Washington Week 30 □ [27]Progress 31 □ [28]Newsletters 32 □ [29][nav-arch]Explore The Atlantic Archive 33 □ [30][crosswor]Play The Atlantic crossword 34 35 The Print Edition 36 37 [31][current-issue] 38 [32]Latest Issue[33]Past Issues 39 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40 [34]Give a Gift 41 • 42 Search The Atlantic 43 [37][ ] 44 Quick Links 45 □ [38]Dear Therapist 46 Dear Therapist 47 □ [39]Crossword Puzzle 48 Crossword Puzzle 49 □ [40]Magazine Archive 50 Magazine Archive 51 □ [41]Your Subscription 52 Your Subscription 53 • [43]Popular 54 • [44]Latest 55 • [45]Newsletters 56 57 [46][47] 58 59 • [48]Sign In 60 • [49]Subscribe 61 62 [50]More From Artificial Intelligence 63 64 More From Artificial Intelligence 65 66 [51]Explore This Series 67 68 • [52]GIF of a water cooler attached to the top of an old desktop computer 69 70 AI Is Taking Water From the Desert 71 72 [53]Karen Hao 73 • [54]Animation of a document being scanned and copied 74 75 Generative AI Is Challenging a 234-Year-Old Law 76 77 [55]Alex Reisner 78 • [56]An image of a Nazi soldier overlaid with a mosaic of brown tiles 79 80 The Deeper Problem With Google’s Racially Diverse Nazis 81 82 [57]Chris Gilliard 83 • [58]A diver descends toward the head of a sperm whale swimming 84 perpendicular to the surface. 85 86 How First Contact With Whale Civilization Could Unfold 87 88 [59]Ross Andersen 89 90 [60]Technology 91 92 Neal Stephenson’s Most Stunning Prediction 93 94 The sci-fi legend coined the term metaverse. But he was most prescient about 95 our AI age. 96 97 98 By [61]Matteo Wong 99 Sci-fi author Neal Stephenson in black and white next to an arm holding a book 100 Illustration by The Atlantic. Sources: Heritage Images; Amy E. Price / Getty. 101 February 6, 2024 102 Share 103 Save 104 105 Science fiction, when revisited years later, sometimes doesn’t come across as 106 all that fictional. Speculative novels have an impressive track record at 107 prophesying what innovations are to come, and how they might upend the world: 108 H. G. Wells wrote about an atomic bomb [64]decades before World War II, and Ray 109 Bradbury’s 1953 novel, Fahrenheit 451, features devices we’d describe today as 110 Bluetooth earbuds. 111 112 Perhaps no writer has been more clairvoyant about our current technological age 113 than Neal Stephenson. His novels coined the term [65]metaverse, laid the 114 conceptual groundwork for cryptocurrency, and imagined a geoengineered planet. 115 And nearly three decades before the release of ChatGPT, he presaged the current 116 AI revolution. A core element of one of his early novels, [66]The Diamond Age: 117 Or, a Young Lady’s Illustrated Primer, is a magical book that acts as a 118 personal tutor and mentor for a young girl, adapting to her learning style—in 119 essence, it is a personalized and ultra-advanced chatbot. The titular Primer 120 speaks aloud in the voice of a live actor, known as a “ractor”—evoking how 121 today’s generative AI, like many digital technologies, is highly dependent on 122 humans’ creative labor. 123 124 Stephenson’s book, published in 1995, explores a future of seamless, instant 125 digital communication, in which tiny computers with immense capabilities are 126 embedded in everyday life. Corporations are dominant, news and ads are 127 targeted, and screens are omnipresent. It’s a world of stark class and cultural 128 divisions (the novel follows a powerful aristocratic sect that styles itself as 129 the “neo-Victorians”), but it’s nevertheless one in which the Primer is 130 presented as the best of what technology can be. 131 132 [67][original] 133 [68]The Diamond Age - Or, A Young Lady's Illustrated Primer 134 By Neal Stephenson 135 Buy Book 136 137 But Stephenson is far more pessimistic about today’s AI than he was about the 138 Primer. “A chatbot is not an oracle,” he told me over Zoom last Friday. “It’s a 139 statistics engine that creates sentences that sound accurate.” I spoke with 140 Stephenson about his uncannily prescient book and the generative-AI revolution 141 that has seemingly begun. 142 143 This conversation has been edited for length and clarity. 144 145 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 146 147 Matteo Wong: The Young Lady’s Illustrated Primer is a book that adapts to and 148 teaches a young girl, which seems to resonate with the vision of AI chatbots 149 and assistants that many companies have for the near future. Did you set out to 150 explore the idea of an intelligent machine in imagining the Primer? 151 152 Neal Stephenson: The idea came to me after we had a kid and got this mobile 153 that was designed to suspend over the crib. It had very primitive, simple 154 shapes on it because, when they’re newborns, their visual systems can’t resolve 155 fine details. So there would be a square and a triangle and a circle. And then, 156 after a certain number of days or weeks had gone by, you were supposed to pop 157 those cards off of the mobile and snap on a different set that had a more 158 appropriate fit for what their brains were capable of at that age. That just 159 got me to thinking: What if you extended that idea to every other form of 160 intellectual growth? 161 162 The technology that drives the book wasn’t really AI as we think of it now—I 163 was talking to people who were working on some of the underlying technologies 164 that would be needed to communicate on the internet in a secure, anonymous 165 manner. I guess it’s implicit that there’s an AI in there that’s generating the 166 story and increasing the degree of sophistication in response to the learning 167 curve of the child, but I didn’t really go into that very much; I just kind of 168 assumed it would be there. 169 170 Wong: A lot of companies today—OpenAI, Google, Meta, to name a few—have said 171 they want to build AI assistants that adapt to each user, somewhat like how the 172 Primer acts as a teacher. Do you see anything in the generative-AI models of 173 today that resembles or could one day become like the Primer? 174 175 Stephenson: About a year ago, I worked with a start-up that makes AI characters 176 in video games. I found it rewarding and fascinating because of the 177 hallucinations: I could see how new patterns emerged from the soup of inputs 178 being fed to it. The same thing that I consider to be a feature is a bug in 179 most applications. We’ve already seen examples of lawyers who use ChatGPT to 180 create legal documents, and the AI just fabricated past cases and precedents 181 that seemed completely plausible. When you think about the idea of trying to 182 make use of these models in education, this becomes a bug too. What they do is 183 generate sentences that sound like correct sentences, but there’s no underlying 184 brain that can actually discern whether those sentences are correct or not. 185 186 [70]Read: The end of high-school English 187 188 Think about any concept that we might want to teach somebody—for instance, the 189 Pythagorean theorem. There must be thousands of old and new explanations of the 190 Pythagorean theorem online. The real thing we need is to understand each 191 child’s learning style so we can immediately connect them to the one out of 192 those thousands that is the best fit for how they learn. That to me sounds like 193 an AI kind of project, but it’s a different kind of AI application from DALL-E 194 or large language models. 195 196 Wong: And yet, today, those language models, which fundamentally predict words 197 in a sequence, are being applied to many areas where they have no specialized 198 abilities—GPT-4 for medical diagnosis, Google Bard as a tutor. That reminds me 199 of a term used in the book instead of artificial intelligence, 200 pseudo-intelligence, which many critics of the technology might appreciate 201 today. 202 203 Stephenson: I’d forgotten about that. The running gag of that book was applying 204 Victorian diction and prejudices to high-tech things. What was probably going 205 through my mind was that Victorians would look askance at the term artificial 206 intelligence, because they would be offended by the idea that computers could 207 replace human brains. So they would probably want to bracket the idea as a 208 simulation, or a “pseudo” intelligence, as opposed to the real thing. 209 210 Wong: About a year ago, in an [71]interview with the Financial Times, you 211 called the outputs of generative AI “hollow and uninteresting.” Why was that, 212 and has your assessment changed? 213 214 Stephenson: I suspect that what I had in mind when I was making those remarks 215 was the current state of image-generating technology. There were a few things 216 about that rubbing me the wrong way, the biggest being that they are benefiting 217 from the uncredited work of thousands of real human artists. I’m going to 218 exaggerate slightly, but it seems like one of the first applications of any new 219 technology is making things even shittier for artists. That’s certainly 220 happened with music. These image-generation systems just seemed like that was 221 mechanized and weaponized on an inconceivable scale. 222 223 [72]Read: These 183,000 books are fueling the biggest fight in publishing and 224 tech 225 226 Another part of it was that a lot of people who got excited about this early on 227 just generated huge volumes of material and put them out willy-nilly on the 228 internet. If your only way of making a painting is to actually dab paint 229 laboriously onto a canvas, then the result might be bad or good, but at least 230 it’s the result of a whole lot of micro-decisions you made as an artist. You 231 were exercising editorial judgment with every paint stroke. That is absent in 232 the output of these programs. 233 234 Wong: Even in The Diamond Age, the Primer seems to provide commentary on 235 artists’ labor and tech, which is very relevant to generative AI today. The 236 Primer teaches a girl, but a human actor digitally connected to the book has to 237 voice the text aloud. 238 239 Stephenson: If you’re a conventional actor onstage or in film, you stand in 240 front of a camera, you perform once, and then lots of copies can be made. In 241 the book, I thought it was a pretty positive vision of the future, where we 242 have the technology that would enable voice actors to in effect give live 243 performances on demand, all the time. Even with today’s voice clones, if you 244 break it down to its simplest element, there’s still a human who sat in front 245 of a microphone and provided this material. Although I guess a system like the 246 Primer might not work live; you would probably have some lag—the AI is 247 generating the text and sending it to the ractor, and then the ractor has to 248 read it. 249 250 Wong: And on the scale that some of today’s AI programs operate on, there just 251 wouldn’t be enough people to do it. 252 253 Stephenson: The scenario I was laying out in The Diamond Age is that the 254 ractors are a scarce resource, and so the Primer is more of a luxury product. 255 But eventually, the source code for the book falls into the hands of a man who 256 wants to manufacture it on a massive scale, and there’s not enough money and 257 not enough actors in the world to voice all those books, so at that point, he 258 decides to use automatically generated voices. 259 260 Wong: Another theme in the novel is how different socioeconomic classes have 261 access to education. The Primer is designed for an aristocrat, but your novel 262 also traces the stories of middle- and working-class girls who interact with 263 versions of the book. Right now a lot of generative AI is free, but the 264 technology is also very expensive to run. How do you think access to generative 265 AI might play out? 266 267 Stephenson: There was a bit of early internet utopianism in the book, which was 268 written during that era in the mid-’90s when the internet was coming online. 269 There was a tendency to assume that when all the world’s knowledge comes 270 online, everyone will flock to it. It turns out that if you give everyone 271 access to the Library of Congress, what they do is watch videos on TikTok. The 272 Diamond Age reflects the same naivete that I shared with a lot of other people 273 back in the day about how all of that knowledge was going to affect society. 274 275 Wong: Do you think we’re seeing some of that naivete today in people looking at 276 how generative AI can be used? 277 278 Stephenson: For sure. It’s based on an understandable misconception as to what 279 these things are doing. A chatbot is not an oracle; it’s a statistics engine 280 that creates sentences that sound accurate. Right now my sense is that it’s 281 like we’ve just invented transistors. We’ve got a couple of consumer products 282 that people are starting to adopt, like the transistor radio, but we don’t yet 283 know how the transistor will transform society. We’re in the transistor-radio 284 stage of AI. I think a lot of the ferment that’s happening right now in the 285 industry is venture capitalists putting money into business plans, and teams 286 that are rapidly evaluating a whole lot of different things that could be done 287 well. I’m sure that some things are going to emerge that I wouldn’t dare try to 288 predict, because the results of the creative frenzy of millions of people are 289 always more interesting than what a single person can think of. 290 291 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 292 293 When you buy a book using a link on this page, we receive a commission. Thank 294 you for supporting The Atlantic. 295 296 297 [73]Matteo Wong is an associate editor at The Atlantic. 298 299 300 References: 301 302 [1] https://www.theatlantic.com/technology/archive/2024/02/chatbots-ai-neal-stephenson-diamond-age/677364/#main-content 303 [2] https://www.theatlantic.com/ 304 [5] https://www.theatlantic.com/most-popular/ 305 [6] https://www.theatlantic.com/latest/ 306 [7] https://www.theatlantic.com/newsletters/ 307 [8] https://www.theatlantic.com/politics/ 308 [9] https://www.theatlantic.com/ideas/ 309 [10] https://www.theatlantic.com/category/fiction/ 310 [11] https://www.theatlantic.com/technology/ 311 [12] https://www.theatlantic.com/science/ 312 [13] https://www.theatlantic.com/photo/ 313 [14] https://www.theatlantic.com/business/ 314 [15] https://www.theatlantic.com/culture/ 315 [16] https://www.theatlantic.com/projects/planet/ 316 [17] https://www.theatlantic.com/international/ 317 [18] https://www.theatlantic.com/books/ 318 [19] https://www.theatlantic.com/podcasts/ 319 [20] https://www.theatlantic.com/health/ 320 [21] https://www.theatlantic.com/education/ 321 [22] https://www.theatlantic.com/projects/ 322 [23] https://www.theatlantic.com/category/features/ 323 [24] https://www.theatlantic.com/family/ 324 [25] https://www.theatlantic.com/events/ 325 [26] https://www.theatlantic.com/category/washington-week-atlantic/ 326 [27] https://www.theatlantic.com/progress/ 327 [28] https://www.theatlantic.com/newsletters/ 328 [29] https://www.theatlantic.com/archive/ 329 [30] https://www.theatlantic.com/free-daily-crossword-puzzle/ 330 [31] https://www.theatlantic.com/magazine/ 331 [32] https://www.theatlantic.com/magazine/ 332 [33] https://www.theatlantic.com/magazine/backissues/ 333 [34] https://accounts.theatlantic.com/products/gift 334 [38] https://www.theatlantic.com/projects/dear-therapist/ 335 [39] https://www.theatlantic.com/free-daily-crossword-puzzle/ 336 [40] https://www.theatlantic.com/archive/ 337 [41] https://accounts.theatlantic.com/accounts/subscription/ 338 [43] https://www.theatlantic.com/most-popular/ 339 [44] https://www.theatlantic.com/latest/ 340 [45] https://www.theatlantic.com/newsletters/ 341 [46] https://www.theatlantic.com/ 342 [47] https://www.theatlantic.com/ 343 [48] https://accounts.theatlantic.com/login/ 344 [49] https://www.theatlantic.com/subscribe/navbar/ 345 [50] https://www.theatlantic.com/projects/artificial-intelligence/ 346 [51] https://www.theatlantic.com/projects/artificial-intelligence/ 347 [52] https://www.theatlantic.com/technology/archive/2024/03/ai-water-climate-microsoft/677602/ 348 [53] https://www.theatlantic.com/author/karen-hao/ 349 [54] https://www.theatlantic.com/technology/archive/2024/02/generative-ai-lawsuits-copyright-fair-use/677595/ 350 [55] https://www.theatlantic.com/author/alex-reisner/ 351 [56] https://www.theatlantic.com/technology/archive/2024/02/google-gemini-diverse-nazis/677575/ 352 [57] https://www.theatlantic.com/author/chris-gilliard/ 353 [58] https://www.theatlantic.com/science/archive/2024/02/talking-whales-project-ceti/677549/ 354 [59] https://www.theatlantic.com/author/ross-andersen/ 355 [60] https://www.theatlantic.com/technology/ 356 [61] https://www.theatlantic.com/author/matteo-wong/ 357 [64] https://thebulletin.org/virtual-tour/h-g-wells-novel-the-world-set-free-predicts-atomic-warfare/ 358 [65] https://www.theatlantic.com/technology/archive/2021/10/facebook-metaverse-name-change/620449/ 359 [66] https://bookshop.org/p/books/the-diamond-age-or-a-young-lady-s-illustrated-primer-neal-stephenson/8466804?ean=9780553380965 360 [67] https://bookshop.org/a/12476/9780553380965 361 [68] https://bookshop.org/a/12476/9780553380965 362 [70] https://www.theatlantic.com/technology/archive/2022/12/openai-chatgpt-writing-high-school-english-essay/672412/ 363 [71] https://www.ft.com/content/0ecab009-6543-4386-b936-0eecc9293d2e 364 [72] https://www.theatlantic.com/technology/archive/2023/09/books3-database-generative-ai-training-copyright-infringement/675363/ 365 [73] https://www.theatlantic.com/author/matteo-wong/