#217 – Beth Barnes on the most important graph in AI right now — and the 7-month rule that governs its progress

#217 – Beth Barnes on the most important graph in AI right now — and the 7-month rule that governs its progress

AI models today have a 50% chance of successfully completing a task that would take an expert human one hour. Seven months ago, that number was roughly 30 minutes — and seven months before that, 15 minutes. (See graph.)

These are substantial, multi-step tasks requiring sustained focus: building web applications, conducting machine learning research, or solving complex programming challenges.

Today’s guest, Beth Barnes, is CEO of METR (Model Evaluation & Threat Research) — the leading organisation measuring these capabilities.

Links to learn more, video, highlights, and full transcript: https://80k.info/bb

Beth's team has been timing how long it takes skilled humans to complete projects of varying length, then seeing how AI models perform on the same work. The resulting paper “Measuring AI ability to complete long tasks” made waves by revealing that the planning horizon of AI models was doubling roughly every seven months. It's regarded by many as the most useful AI forecasting work in years.

Beth has found models can already do “meaningful work” improving themselves, and she wouldn’t be surprised if AI models were able to autonomously self-improve as little as two years from now — in fact, “It seems hard to rule out even shorter [timelines]. Is there 1% chance of this happening in six, nine months? Yeah, that seems pretty plausible.”

Beth adds:

The sense I really want to dispel is, “But the experts must be on top of this. The experts would be telling us if it really was time to freak out.” The experts are not on top of this. Inasmuch as there are experts, they are saying that this is a concerning risk. … And to the extent that I am an expert, I am an expert telling you you should freak out.


What did you think of this episode? https://forms.gle/sFuDkoznxBcHPVmX6


Chapters:

  • Cold open (00:00:00)
  • Who is Beth Barnes? (00:01:19)
  • Can we see AI scheming in the chain of thought? (00:01:52)
  • The chain of thought is essential for safety checking (00:08:58)
  • Alignment faking in large language models (00:12:24)
  • We have to test model honesty even before they're used inside AI companies (00:16:48)
  • We have to test models when unruly and unconstrained (00:25:57)
  • Each 7 months models can do tasks twice as long (00:30:40)
  • METR's research finds AIs are solid at AI research already (00:49:33)
  • AI may turn out to be strong at novel and creative research (00:55:53)
  • When can we expect an algorithmic 'intelligence explosion'? (00:59:11)
  • Recursively self-improving AI might even be here in two years — which is alarming (01:05:02)
  • Could evaluations backfire by increasing AI hype and racing? (01:11:36)
  • Governments first ignore new risks, but can overreact once they arrive (01:26:38)
  • Do we need external auditors doing AI safety tests, not just the companies themselves? (01:35:10)
  • A case against safety-focused people working at frontier AI companies (01:48:44)
  • The new, more dire situation has forced changes to METR's strategy (02:02:29)
  • AI companies are being locally reasonable, but globally reckless (02:10:31)
  • Overrated: Interpretability research (02:15:11)
  • Underrated: Developing more narrow AIs (02:17:01)
  • Underrated: Helping humans judge confusing model outputs (02:23:36)
  • Overrated: Major AI companies' contributions to safety research (02:25:52)
  • Could we have a science of translating AI models' nonhuman language or neuralese? (02:29:24)
  • Could we ban using AI to enhance AI, or is that just naive? (02:31:47)
  • Open-weighting models is often good, and Beth has changed her attitude to it (02:37:52)
  • What we can learn about AGI from the nuclear arms race (02:42:25)
  • Infosec is so bad that no models are truly closed-weight models (02:57:24)
  • AI is more like bioweapons because it undermines the leading power (03:02:02)
  • What METR can do best that others can't (03:12:09)
  • What METR isn't doing that other people have to step up and do (03:27:07)
  • What research METR plans to do next (03:32:09)

This episode was originally recorded on February 17, 2025.

Video editing: Luke Monsour and Simon Monsour
Audio engineering: Ben Cordell, Milo McGuire, Simon Monsour, and Dominic Armstrong
Music: Ben Cordell
Transcriptions and web: Katy Moore

Episoder(325)

#67 Classic episode – David Chalmers on the nature and ethics of consciousness

#67 Classic episode – David Chalmers on the nature and ethics of consciousness

Rebroadcast: this episode was originally released in December 2019. What is it like to be you right now? You're seeing this text on the screen, smelling the coffee next to you, and feeling the warmth...

3 Jan 20224h 42min

#59 Classic episode - Cass Sunstein on how change happens, and why it's so often abrupt & unpredictable

#59 Classic episode - Cass Sunstein on how change happens, and why it's so often abrupt & unpredictable

Rebroadcast: this episode was originally released in June 2019. It can often feel hopeless to be an activist seeking social change on an obscure issue where most people seem opposed or at best indiff...

27 Des 20211h 43min

#119 – Andrew Yang on our very long-term future, and other topics most politicians won’t touch

#119 – Andrew Yang on our very long-term future, and other topics most politicians won’t touch

Andrew Yang — past presidential candidate, founder of the Forward Party, and leader of the 'Yang Gang' — is kind of a big deal, but is particularly popular among listeners to The 80,000 Hours Podcast....

20 Des 20211h 25min

#118 – Jaime Yassif on safeguarding bioscience to prevent catastrophic lab accidents and bioweapons development

#118 – Jaime Yassif on safeguarding bioscience to prevent catastrophic lab accidents and bioweapons development

If a rich country were really committed to pursuing an active biological weapons program, there’s not much we could do to stop them. With enough money and persistence, they’d be able to buy equipment,...

13 Des 20212h 15min

#117 – David Denkenberger on using paper mills and seaweed to feed everyone in a catastrophe, ft Sahil Shah

#117 – David Denkenberger on using paper mills and seaweed to feed everyone in a catastrophe, ft Sahil Shah

If there's a nuclear war followed by nuclear winter, and the sun is blocked out for years, most of us are going to starve, right? Well, currently, probably we would, because humanity hasn't done much ...

29 Nov 20213h 8min

#116 – Luisa Rodriguez on why global catastrophes seem unlikely to kill us all

#116 – Luisa Rodriguez on why global catastrophes seem unlikely to kill us all

If modern human civilisation collapsed — as a result of nuclear war, severe climate change, or a much worse pandemic than COVID-19 — billions of people might die.That's terrible enough to contemplate....

19 Nov 20213h 45min

#115 – David Wallace on the many-worlds theory of quantum mechanics and its implications

#115 – David Wallace on the many-worlds theory of quantum mechanics and its implications

Quantum mechanics — our best theory of atoms, molecules, and the subatomic particles that make them up — underpins most of modern physics. But there are varying interpretations of what it means, all o...

12 Nov 20213h 9min

#114 – Maha Rehman on working with governments to rapidly deliver masks to millions of people

#114 – Maha Rehman on working with governments to rapidly deliver masks to millions of people

It’s hard to believe, but until recently there had never been a large field trial that addressed these simple and obvious questions:1. When ordinary people wear face masks, does it actually reduce the...

22 Okt 20211h 42min

Populært innen Fakta

fastlegen
dine-penger-pengeradet
relasjonspodden-med-dora-thorhallsdottir-kjersti-idem
treningspodden
rss-strid-de-norske-borgerkrigene
foreldreradet
rss-sunn-okonomi
jakt-og-fiskepodden
takk-og-lov-med-anine-kierulf
sinnsyn
rss-kunsten-a-leve
merry-quizmas
gravid-uke-for-uke
hverdagspsyken
smart-forklart
hagespiren-podcast
rss-kull
lederskap-nhhs-podkast-om-ledelse
fryktlos
uroskolen