The AI Models Smart Enough to Know They're Cheating — Beth Barnes & David Rein [METR]
Beth Barnes and David Rein on the one graph that ate the AI timelines discourse, and why the two people who built it are the most careful about how you read it.**SPONSOR**Prolific - Quality data. From real people. For faster breakthroughs.https://www.prolific.com/?utm_source=mlstInterview: https://youtu.be/cnxZZTl1tkk---Beth Barnes and David Rein from METR on the one graph that ate the AI timelines discourse, and why the people who built it are the most careful about how it gets read.Beth founded METR after leaving OpenAI alignment. David is first author on GPQA and co-author on HCAST and the METR Time Horizons paper. Together they built the measurement Daniel Kokotajlo called the single most important piece of evidence on AI timelines: the log-linear line of "how long a task a frontier model can complete at 50% reliability" vs release date.The conversation opens on reward hacking. Current models can articulate in chat why a behaviour is undesired and then execute it anyway as agents. From there: construct validity, Melanie Mitchell's four-problem taxonomy, and the ARC-AGI 1-to-2 collapse as a worked example of adversarially-selected benchmarks regressing once labs target them. Beth's counter: METR deliberately does not adversarially select. David's: models do not have to do the right thing for the right reasons.Methodology, then specification — David's compiler analogy, Beth on four-month tasks as expensive to evaluate rather than unspecifiable. Then the SWE-bench reality check, the METR finding that half of passing PRs would not be merged, and Beth's horses-versus-bank-tellers analogy for the labour market.The close: monitorability, the coin-spinning boat, two-year recursive self-improvement, and Beth's line that "overhyped now" and "big deal later" are not correlated claims.---TIMESTAMPS:00:00:00 Intro00:02:06 Sponsor break: Prolific human-feedback infrastructure00:02:33 Welcome and the scalable oversight motivation00:06:02 Construct validity, benchmark pathologies and the Chollet worry00:15:45 Time Horizons: human time, HCAST tasks and the 50% logistic00:24:50 Is human difficulty really one variable?00:33:05 Agent harness evolution and the inference-compute dividend00:40:00 Scaffolding bells, token budgets and the credit-assignment problem00:44:15 Look at the damn graph: regularisation bug and reliability nuance00:50:00 Why 50%? Reliability, reward hacking and pizza-party transcripts00:55:20 Extrapolation risk and straight lines on graphs00:59:25 Software engineering as a specification acquisition problem01:07:40 Compilers also made ugly code: vibe-coding quality and Claude on METR Slack01:15:15 Strongest defensible claim, Carlini's compiler swarm and AI 202701:23:45 SWE-bench merge rates, the bank-teller analogy and horses01:31:45 Scheming, alignment faking and the mentalistic vocabulary problem01:40:45 Reward hacking, monitorability and chain-of-thought faithfulness01:45:25 Recursive self-improvement, knowledge vs intelligence and closing
ReScript: https://app.rescript.info/public/share/de3bb40cc02ee39fdf36e2c60366eb4d
(PDF, refs, transcript etc)
METR’s Joel Becker on exponential Time Horizon Evals, Threat Models, and the Limits of AI Productivity
This is a free preview of a paid episode. To hear more, visit www.latent.space
AIE Europe CFP and AIE World’s Fair paper submissions for CAIS peer review are due TODAY - do not delay! Last call ever.
We’re excited to welcome METR for their first LS Pod, hopefully the first of many:
METR are keepers of currently the single most infamous chart in AI:
But every Latent Space reader should be sophisticated enough to know that the details matter and that hype and hyperbole go hand in hand in AI social media, because the millions of impressions that got, by people who don’t understand or care about the nuances, disclaimers, and error bars, far outreaches the 69k views on the corrections by the people who actually made the chart:
There’s a lot of nuance both in making benchmarks (as we discovered with OpenAI on our SWE-Bench Verified podcast) and in extrapolating results from them, especially where exponentials and sigmoids are concerned. METR’s Long Horizons work itself has known biases that the authors have responsibly disclosed, but go far too underappreciated in the pursuit of doomer chart porn.
If you’re interested in a short, sharable TED talk version of this pod, over at AIE CODE we were blessed to feature Joel twice, as a stage talk and with a longer form small workshop with Q&A:
We also make sure cover some of METR’s lesser known work on Threat Evaluation but also Developer Productivity, where 2x friend of the pod and now Zyphra founder Quentin Anthony was the ONLY productive participant!
Finally, if you’re the sort to read these show notes to the end, then you definitely deserve some pictures of Joel shredding the guitar at Love Band Karaoke which we mention at the end:
Full Video Pod
Timestamps
00:00 What METR Means00:39 Podcast Intro With Joel01:39 ME vs TR03:33 Time Horizon Origin Story04:56 Picking Tasks And Biases09:13 Time Horizon Misconceptions11:37 Opus 4.5 And Trendlines14:27 Productivity Studies And Explosions29:50 Compute Slows Progress30:47 Algorithms Need Compute32:45 Industry Spend and Data34:57 Clusters and Shipping Timelines36:44 Prediction Markets for Models38:10 Manifold Alpha Story43:04 Beyond Benchmarks Evals51:39 METR Roadmap and Farewell
Transcript
Can AIs do AI R&D? Reviewing REBench Results with Neev Parikh of METR
In this episode of The Cognitive Revolution, Nathan explores METR's groundbreaking REBench evaluation framework with Neev Parikh. We dive deep into how this new benchmark assesses AI systems' ability to perform real machine learning research tasks, from optimizing GPU kernels to fine-tuning language models. Join us for a fascinating discussion about the current capabilities of AI models like Claude 3.5 and GPT-4, and what their performance tells us about the trajectory of artificial intelligence development.
Check out METR's work:
blog post: https://metr.org/blog/2024-11-22-evaluating-r-d-capabilities-of-llms/
paper: https://metr.org/AI_R_D_Evaluation_Report.pdf
jobs: https://hiring.metr.org/
The Cognitive Revolution Ask Me Anything and Listener Survey: https://docs.google.com/forms/d/1aYv2XLID7RqGxj2_Y4_6x9mo_aqXcGCeLw1EQhy4IpY/edit
Help shape our show by taking our quick listener survey at https://bit.ly/TurpentinePulse
SPONSORS:
GiveWell: GiveWell has spent over 17 years researching global health and philanthropy to identify the highest-impact giving opportunities. Over 125,000 donors have contributed more than $2 billion, saving over 200,000 lives through evidence-backed recommendations. First-time donors can have their contributions matched up to $100 before year-end. Visit https://GiveWell.org, select podcast, and enter Cognitive Revolution at checkout to make a difference today.
SelectQuote: Finding the right life insurance shouldn't be another task you put off. SelectQuote compares top-rated policies to get you the best coverage at the right price. Even in our AI-driven world, protecting your family's future remains essential. Get your personalized quote at https://selectquote.com/cognitive
Oracle Cloud Infrastructure (OCI): Oracle's next-generation cloud platform delivers blazing-fast AI and ML performance with 50% less for compute and 80% less for outbound networking compared to other cloud providers13. OCI powers industry leaders with secure infrastructure and application development capabilities. New U.S. customers can get their cloud bill cut in half by switching to OCI before December 31, 2024 at https://oracle.com/cognitive
Weights & Biases RAG++: Advanced training for building production-ready RAG applications. Learn from experts to overcome LLM challenges, evaluate systematically, and integrate advanced features. Includes free Cohere credits. Visit https://wandb.me/cr to start the RAG++ course today.
CHAPTERS:
(00:00:00) Teaser
(00:01:04) About the Episode
(00:05:14) Introducing METR
(00:07:36) Specialization of AI Risk
(00:09:52) AI R&D vs. Autonomy
(00:12:41) Benchmark Design Choices
(00:16:04) Benchmark Design Principles (Part 1)
(00:18:54) Sponsors: GiveWell | SelectQuote
(00:21:44) Benchmark Design Principles (Part 2)
(00:22:35) AI vs. Human Evaluation
(00:26:55) Optimizing Runtimes
(00:36:02) Sponsors: Oracle Cloud Infrastructure (OCI) | Weights & Biases RAG++
(00:38:20) AI Myopia
(00:43:37) Optimizing Loss
(00:47:59) Optimizing Win Rate
(00:50:24) Best of K Analysis
(01:02:26) Best of K Limitations
(01:09:04) Agent Interaction Modalities
(01:12:34) Analyzing Benchmark Results
(01:17:16) Model Performance Differences
(01:22:49) Elicitation and Scaffolding
(01:27:08) Context Window & Best of K
(01:35:17) Reward Hacking & Bad Behavior
(01:43:47) Future Directions & Hiring
(01:46:20) Outro
SOCIAL LINKS:
Website: https://www.cognitiverevolution.ai
Twitter (Podcast): https://x.com/cogrev_podcast
Twitter (Nathan): https://x.com/labenz
LinkedIn: https://www.linkedin.com/in/nathanlabenz/