Podcasts

Filtered to Benchmark limitations · clear filter

Browse by topic

Reinforcement learning 11AI agents 10recursive self-improvement 7AI safety 6Diffusion models 5formal verification 5Agentic AI 4AI ethics 3AI infrastructure 3Code generation 3Human-AI collaboration 3Mechanistic interpretability 3multi-agent systems 3Open Source AI 3Scaling laws 3Synthetic data generation 3Venture capital 3world models 3Agent architecture 2Computational complexity 2Context engineering 2Continual learning 2Continuous learning 2Data generation 2Developer experience 2Drug discovery 2Foundation models 2GPU infrastructure 2Lab automation 2LLM inference 2

Matching episodes

The AI Models Smart Enough to Know They're Cheating — Beth Barnes & David Rein [METR]

Machine Learning Street Talk (MLST) · 4 MAY 2026 · 113 min

This episode features Beth Barnes and David Rein from METR discussing their 'Time Horizon' graph, a unified metric for measuring AI progress based on human task completion time. They explain how this benchmark addresses the limitati…

AI evaluationScalable oversightBenchmark limitationsData contaminationReward hackingTime Horizon metricHuman baseliningAgentic AIAI capabilitiesAI timelinesSoftware engineering automationAI riskDeceptive alignmentRecursive self-improvementAI monitoring