#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

Jaksot(321)

#173 – Jeff Sebo on digital minds, and how to avoid sleepwalking into a major moral catastrophe

#173 – Jeff Sebo on digital minds, and how to avoid sleepwalking into a major moral catastrophe

"We do have a tendency to anthropomorphise nonhumans — which means attributing human characteristics to them, even when they lack those characteristics. But we also have a tendency towards anthropoden...

22 Marras 20232h 38min

#172 – Bryan Caplan on why you should stop reading the news

#172 – Bryan Caplan on why you should stop reading the news

Is following important political and international news a civic duty — or is it our civic duty to avoid it?It's common to think that 'staying informed' and checking the headlines every day is just wha...

17 Marras 20232h 23min

#171 – Alison Young on how top labs have jeopardised public health with repeated biosafety failures

#171 – Alison Young on how top labs have jeopardised public health with repeated biosafety failures

"Rare events can still cause catastrophic accidents. The concern that has been raised by experts going back over time, is that really, the more of these experiments, the more labs, the more opportunit...

9 Marras 20231h 46min

#170 – Santosh Harish on how air pollution is responsible for ~12% of global deaths — and how to get that number down

#170 – Santosh Harish on how air pollution is responsible for ~12% of global deaths — and how to get that number down

"One [outrageous example of air pollution] is municipal waste burning that happens in many cities in the Global South. Basically, this is waste that gets collected from people's homes, and instead of ...

1 Marras 20232h 57min

#169 – Paul Niehaus on whether cash transfers cause economic growth, and keeping theft to acceptable levels

#169 – Paul Niehaus on whether cash transfers cause economic growth, and keeping theft to acceptable levels

"One of our earliest supporters and a dear friend of mine, Mark Lampert, once said to me, “The way I think about it is, imagine that this money were already in the hands of people living in poverty. I...

26 Loka 20231h 47min

#168 – Ian Morris on whether deep history says we're heading for an intelligence explosion

#168 – Ian Morris on whether deep history says we're heading for an intelligence explosion

"If we carry on looking at these industrialised economies, not thinking about what it is they're actually doing and what the potential of this is, you can make an argument that, yes, rates of growth a...

23 Loka 20232h 43min

#167 – Seren Kell on the research gaps holding back alternative proteins from mass adoption

#167 – Seren Kell on the research gaps holding back alternative proteins from mass adoption

"There have been literally thousands of years of breeding and living with animals to optimise these kinds of problems. But because we're just so early on with alternative proteins and there's so much ...

18 Loka 20231h 54min

#166 – Tantum Collins on what he’s learned as an AI policy insider at the White House, DeepMind and elsewhere

#166 – Tantum Collins on what he’s learned as an AI policy insider at the White House, DeepMind and elsewhere

"If you and I and 100 other people were on the first ship that was going to go settle Mars, and were going to build a human civilisation, and we have to decide what that government looks like, and we ...

12 Loka 20233h 8min

Suosittua kategoriassa Koulutus

rss-murhan-anatomia
psykopodiaa-podcast
voi-hyvin-meditaatiot-2
rss-narsisti
rss-liian-kuuma-peruna
rss-vapaudu-voimaasi
dear-ladies
aamukahvilla
psykologia
leveli
kesken
ihminen-tavattavissa-tommy-hellsten-instituutti
rss-uskonto-on-tylsaa
rss-duodecim-lehti
rss-valo-minussa-2
rahapuhetta
adhd-podi
rss-tietoinen-yhteys-podcast-2
rss-hereilla
rss-xamk-podcast