Why AI Evaluations Are Broken and How to Fix Them (with David Manheim)

Why AI Evaluations Are Broken and How to Fix Them (with David Manheim)

David Manheim is head of methodology at AI Evaluation Consensus. He joins the podcast to discuss how AI evaluations can become more reliable, transparent, and useful for decisions. We cover common failures such as unclear reporting, training to the test, benchmark saturation, and models changing behavior when they know they are being tested. The conversation also examines real-world tests, biosecurity, persuasion, forecasting, human oversight, and why even “normal” AI progress could be disruptive. LINKS: * David Manheim website [https://davidmanheim.com] * AI Evaluation Consensus Statement [https://evals-consensus.ai/] CHAPTERS: (00:00) Episode Preview (01:04) Evaluation consensus project (07:01) Evaluation awareness challenges (12:28) Reporting capabilities clearly (19:38) Benchmarks beyond humans (29:52) Proxies and biosecurity (42:01) Persuasion and democracy (53:59) Forecasting with AI (01:08:44) Oversight and disruption (01:16:42) Supporting better evals PRODUCED BY: https://aipodcast.ing [https://aipodcast.ing] SOCIAL LINKS: Website: https://podcast.futureoflife.org [https://podcast.futureoflife.org] Twitter (FLI): https://x.com/FLI_org [https://x.com/FLI_org] Twitter (Gus): https://x.com/gusdocker [https://x.com/gusdocker] LinkedIn: https://www.linkedin.com/company/future-of-life-institute/ [https://www.linkedin.com/company/future-of-life-institute/] YouTube: https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/ [https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/] Apple: https://geo.itunes.apple.com/us/podcast/id1170991978 [https://geo.itunes.apple.com/us/podcast/id1170991978] Spotify: https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP [https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP]

How AI Is Replacing Children's Ability to Think (with Randi Weingarten)

Randi Weingarten is president of the American Federation of Teachers. She joins the podcast to discuss AI in education. The conversation covers how screens and student-facing AI may weaken attention, learning, and trust, why she supports limiting screens for young children and social chatbots for students under 16, and how teachers can use AI without replacing thinking. She also discusses active learning, data privacy, independent research on AI in education, and guardrails for schools. LINKS: * Randi Weingarten Wikipedia [https://en.wikipedia.org/wiki/Randi_Weingarten] * AFT Randi Weingarten Profile [https://www.aft.org/about/leadership/randi-weingarten] * American Federation of Teachers [https://www.aft.org] CHAPTERS: (00:00) Episode Preview (01:13) AI education wake-up (07:27) Active learning versus AI (14:42) Preserving thinking habits (20:06) Research and screen limits (26:05) Social chatbot risks (30:20) Data privacy standards (35:58) Classroom AI guardrails (41:13) Companies and future schools PRODUCED BY: https://aipodcast.ing [https://aipodcast.ing] SOCIAL LINKS: Website: https://podcast.futureoflife.org [https://podcast.futureoflife.org] Twitter (FLI): https://x.com/FLI_org [https://x.com/FLI_org] Twitter (Gus): https://x.com/gusdocker [https://x.com/gusdocker] LinkedIn: https://www.linkedin.com/company/future-of-life-institute/ [https://www.linkedin.com/company/future-of-life-institute/] YouTube: https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/ [https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/] Apple: https://geo.itunes.apple.com/us/podcast/id1170991978 [https://geo.itunes.apple.com/us/podcast/id1170991978] Spotify: https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP [https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP]

The Gulf Between AI Progress and Political Understanding (with Dex Hunter-Torricke)

Dex Hunter-Torricke is founder of the Centre for Tomorrow. He joins the podcast to discuss why AI governance needs to move beyond technical fixes. The conversation covers how AI companies make decisions, why policymakers often misunderstand agents and economic change, and how automation could reshape jobs, welfare, taxation, and global trade. Dex argues that societies need broader political planning before a crisis drives rushed choices. LINKS: * Dex Hunter-Torricke Website [https://dexhuntertorricke.com] * Center for Tomorrow Website [https://www.centerfortomorrow.com] CHAPTERS: (00:00) Episode Preview (01:09) Dex's technology journey (08:22) Broken AI governance (18:00) Big tech blindspots (27:36) Employee power shifts (37:18) Global views and UBI (48:17) Power and displacement (01:00:54) Reasons for optimism PRODUCED BY: https://aipodcast.ing [https://aipodcast.ing] SOCIAL LINKS: Website: https://podcast.futureoflife.org [https://podcast.futureoflife.org] Twitter (FLI): https://x.com/FLI_org [https://x.com/FLI_org] Twitter (Gus): https://x.com/gusdocker [https://x.com/gusdocker] LinkedIn: https://www.linkedin.com/company/future-of-life-institute/ [https://www.linkedin.com/company/future-of-life-institute/] YouTube: https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/ [https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/] Apple: https://geo.itunes.apple.com/us/podcast/id1170991978 [https://geo.itunes.apple.com/us/podcast/id1170991978] Spotify: https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP [https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP]

How AI Companions Trap Users Through Addictive Design (with Claire Boine)

Claire Boine is an assistant professor in technology, law, and AI governance at the European University Institute. She joins the podcast to discuss AI companions and human attachment. The conversation examines how design choices and free-to-start business models can foster dependency, expose intimate data, and blur the lines between therapy, romance, and manipulation. We also cover risks for children and teens, gaps in EU and US law, and why AI policy should distinguish technical alignment from incentives and social harms. LINKS: * Claire Boine Website [https://www.claireboine.com] * Successif Organization Website [https://www.successif.org] CHAPTERS: (00:00) Episode Preview (01:06) Introducing Claire Boine (02:05) Companion app designs (06:23) Feelings and freemium (14:28) Users, attachment, harms (25:06) Age-based risks (35:18) Future relationship norms (41:05) Legal loopholes today (44:19) Rethinking alignment framing (57:19) Narratives and fiduciaries (01:05:47) Collective policy action PRODUCED BY: https://aipodcast.ing [https://aipodcast.ing] SOCIAL LINKS: Website: https://podcast.futureoflife.org [https://podcast.futureoflife.org] Twitter (FLI): https://x.com/FLI_org [https://x.com/FLI_org] Twitter (Gus): https://x.com/gusdocker [https://x.com/gusdocker] LinkedIn: https://www.linkedin.com/company/future-of-life-institute/ [https://www.linkedin.com/company/future-of-life-institute/] YouTube: https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/ [https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/] Apple: https://geo.itunes.apple.com/us/podcast/id1170991978 [https://geo.itunes.apple.com/us/podcast/id1170991978] Spotify: https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP [https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP]

Why AI Chatbots Are a Rival to the Family (with Michael Toscano)

Michael Toscano is a Senior Fellow at the Institute for Family Studies and Director of its Family First Technology Initiative. He joins the podcast to discuss family-centered AI policy. The conversation covers AI companions, self-harm risks, sexualized chatbots, education, smartphones in schools, and why "infinite patience" can harm children's growth. Toscano also explains Catholic social teaching, public pushback against rapid AI deployment, and why society-wide governance may be needed to keep technology accountable to families. LINKS: * Michael Toscano [https://ifstudies.org/about-us/michael-toscano] CHAPTERS: (00:00) Episode Preview (01:12) Family centered AI (09:54) Companion design harms (17:23) Technology and family (25:48) Society wide governance (33:09) Catholic AI response (39:44) Double movement politics (49:36) Technology mythos challenged (59:48) Competition and control PRODUCED BY: https://aipodcast.ing [https://aipodcast.ing] SOCIAL LINKS: Website: https://podcast.futureoflife.org [https://podcast.futureoflife.org] Twitter (FLI): https://x.com/FLI_org [https://x.com/FLI_org] Twitter (Gus): https://x.com/gusdocker [https://x.com/gusdocker] LinkedIn: https://www.linkedin.com/company/future-of-life-institute/ [https://www.linkedin.com/company/future-of-life-institute/] YouTube: https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/ [https://www.youtube.com/channel/UC-rCCy3FQ-GItDimSR9lhzw/] Apple: https://geo.itunes.apple.com/us/podcast/id1170991978 [https://geo.itunes.apple.com/us/podcast/id1170991978] Spotify: https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP [https://open.spotify.com/show/2Op1WO3gwVwCrYHg4eoGyP]