翻訳待ち：Foomflops: Three Decades of AI Predictions

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

About this timeline Follow each prediction from its original source through its deadline and resolution. Every prediction and chapter has a permanent, shareable page with citations. Chapters Chapter 01: Extropians mid-1990s – 2000 — When transhumanism met mailing-list accelerationism Chapter 02: SL4 / Plan to Singularity 1999–2000 — Shock Level Four and open-source seed AI Chapter 03: Singularity Institute for Artificial Intelligence 2000–~2012 — From 'we will build Friendly AI' to early catastrophic risk warnings Chapter 04: Overcoming Bias / LessWrong / Rationality community 2006–2012 — Debugging humanity, one very long blog post at a time Chapter 05: MIRI: Research 2013–2021 — those who can't build, align Chapter 06: MIRI: Advocacy 2022–2025 — those who can't align, sabotage Chapter 07: When Prophecy Fails 2026 – present — the prophecy recedes; the apparatus grows All predictions Singularity by 2025 (or 2005, given funding!) 1996 → 2001; walked back. “It's expected in 2035. (Oops, make that 2025.)” Seed AI by 2010, to outrace nanotech 1999 → 2010; falsified by events. “To have a good chance of outracing nanotech, we should plan to develop a seed AI by 2010.” Nanotech ~2010; the Singularity ~2020 1999 → 2020; falsified by events. “The 50% point [for nanotech] may be 2010, or earlier. … AI at 2020 CRNS.” Neural nets: “certainly Not Sufficient” 2000 → 2023; falsified by events. “Regardless of whether neural networks are Necessary, they are certainly Not Sufficient.” Hard takeoff in a basement within ten years 2001 → 2011; falsified by events. “…a hard takeoff occurs in SIAI's or Webmind's basement sometime in the next ten years or so.” “Three hours to transhumanity” 2001 → 2008; walked back. “…the ‘hard takeoff’ ‘seed AI’ three-hours-to-transhumanity scenario.” Time to Singularity: “2005–2020, peak 2008–2010” 2001 → 2020; falsified by events. “2005-2020 / peak at 2008-2010” SIAI ships transhuman AI ~2008–2010 2001 → 2010; falsified by events. “…reaching transhumanity sometime between 2005 and 2020, probably around 2008 or 2010.” “Dead certain”: days to nanotech 2002 → 2023; falsified by events. “I would call it dead certain in favor of a hard takeoff… all economic, computational… indicators point straight toward it.” “A planet-sized neural network would fail” 2006 → 2023; falsified by events. “But a neural network the size of a planet, given the same examples, would have failed in the same way.” AGI = ten insights, a small team, one desktop 2008 → 2023; falsified by events. “Intelligence is about architecture… architecture is mostly about deep insights.” Doom by 2030, at 2:1 odds 2017 → 2030; still open. “If humanity survives to 00:00 GMT, Jan 1 2030, Yudkowsky pays Caplan $200; a bet at 2:1 odds in favour of doom.” No GDP boom before doom 2021 → 2025; falsified by events. “I'm happy to argue that there won't be a 4-year interval with world economic output doubling before then.” AI takes IMO gold by 2025 (≥16%) 2022 → 2025; came true. “I'll stand by a >16% probability of the technical capability existing by end of 2025.” “Dying with dignity”: certain doom, no date 2022 → open; still open. “…doubling our chances of survival will only take them from 0% to 0%.” Self-driving cars: approved only after the world ends 2021 → 2025; falsified by events. “…the World-Ending Company will be able to use the same tech to build a true self-driving car, which would in the natural course of things be approved for sale a few years later after the world had ended.” The trillion-scale AI boom: 'pretty darn prohibited' 2021 → 2025; falsified by events. “this all sure does sound "pretty darn prohibited" on my model, but I'd hope there'd be something earlier than that we could bet on.” Deepnets as ubiquitous as computers, maybe by 2035 2021 → 2025; falsified by events. “if the world lasts until 2035, I could perhaps see deepnets becoming as ubiquitous as computers were in… 1995? 2005? would that fulfill the terms of the Prophecy?” Codex: stagnation for 5 years or a massive leap in 2 2021 → 2023; hedged: couldn't lose. “I feel like I would indeed not have the right to feel very surprised if Codex technology stagnated for the next 5 years, nor if it took a massive leap in 2 years and got ubiquitously adopted by lots of programmers.” $1B/yr waifutech before AI freedom-of-the-road 2021 → open; still open. “I think my sheer impression would be that we'll get $1B/yr waifutech before we get AI freedom-of-the-road”