How can AIs know what we want if we don't even know? (with Geoffrey Irving)

from Clearer Thinking with Spencer Greenberg · host Spencer Greenberg

Read the full transcript here. What does it really mean to align an AI system with human values? What would a powerful AI need to do in order to do "what we want"? How does being an assistant differ from being an agent? Could inter-AI debate work as an alignment strategy, or would it just result in arguments designed to manipulate humans via their cognitive and emotional biases? How can we make sure that all human values are learned by AIs, not just the values of humans in WEIRD societies? Are our current state-of-the-art LLMs politically left-leaning? How can alignment strategies take into account the fact that our individual and collective values occasionally change over time? Geoffrey Irving is an AI safety researcher at DeepMind. Before that, he led the Reflection Team at OpenAI, was involved in neural network theorem proving at Google Brain, cofounded Eddy Systems to autocorrect code as you type, and worked on computational physics and geometry at Otherlab, D. E. Shaw Research, Pixar, and Weta Digital. He has screen credits on Ratatouille, WALL•E, Up, and Tintin. Learn more about him at his website, naml.us. Further reading Gandalf: An Educational Game Demonstrating Security Vulnerabilities in Large Language Models "AI safety via debate" "Claude's Constitution" Staff Spencer Greenberg — Host / Director Josh Castle — Producer Ryan Kessler — Audio Engineer Uri Bram — Factotum WeAmplify — Transcriptionists Music Broke for Free Josh Woodward Lee Rosevere Quiet Music for Tiny Robots wowamusic zapsplat.com Affiliates Clearer Thinking GuidedTrack Mind Ease Positly UpLift [Read more]

What this episode covers

Read the full transcript here. • What does it really mean to align an AI system with human values? What would a powerful AI need to do in order to do "what we want"? How does being an assistant differ from being an agent? Could inter-AI debate work as an alignment strategy, or would it just result in arguments designed to manipulate humans via their cognitive and emotional biases? How can we make sure that all human values are learned by AIs, not just the values of humans in WEIRD societies? Are our current state-of-the-art LLMs politically left-leaning? How can alignment strategies take into account the fact that our individual and collective values occasionally change over time? • Geoffrey Irving is an AI safety researcher at DeepMind. Before that, he led the Reflection Team at OpenAI, was involved in neural network theorem proving at Google Brain, cofounded Eddy Systems to autocorrect code as you type, and worked on computational physics and geometry at Otherlab, D. E. Shaw Research, Pixar, and Weta Digital. He has screen credits on Ratatouille , WALL•E , Up , and Tintin . Learn more about him at his website, naml.us (https://naml.us/). • Further reading • Gandalf: An Educational Game Demonstrating Security Vulnerabilities in Large Language Models (https://gandalf.lakera.ai/) • "AI safety via debate" (https://openai.com/research/debate) • "Claude's Constitution" (https://www.anthropic.com/index/claudes-constitution) • Staff • Spencer Greenberg (https://www.spencergreenberg.com/) — Host / Director • Josh Castle (mailto:[email protected]) — Producer • Ryan Kessler (https://tone.support/) — Audio Engineer • Uri Bram (https://uribram.com/) — Factotum • WeAmplify (https://www.weamplify.info/) — Transcriptionists • Music • Broke for Free (https://freemusicarchive.org/music/Broke_For_Free/Something_EP/Broke_For_Free_-_Something_EP_-_05_Something_Elated) • Josh Woodward (https://www.joshwoodward.com/song/AlreadyThere) • Lee Rosevere (https://archive.org/details/MusicForPodcasts04/Lee+Rosevere+-+Music+for+Podcasts+4+-+11+Keeping+Stuff+Together.flac) • Quiet Music for Tiny Robots (https://www.freemusicarchive.org/music/Quiet_Music_for_Tiny_Robots/The_February_Album/05_Tiny_Robot_Armies) • wowamusic (https://gamesounds.xyz/?dir=wowamusic) • zapsplat.com (https://www.zapsplat.com/music/summer-haze-slow-chill-out-house-track-with-a-modern-pop-feel-warm-piano-chords-underpin-the-track-with-warm-pads-and-a-repetitive-synth-arpeggio/) • Affiliates • Clearer Thinking (https://www.clearerthinking.org/) • GuidedTrack (https://guidedtrack.com/) • Mind Ease (https://mindease.io/) • Positly (https://positly.com/) • UpLift (https://www.uplift.app/)

NOW PLAYING

How can AIs know what we want if *we* don't even know? (with Geoffrey Irving)

0:00 1:19:53

1×

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

Share this episode

Similar Episodes

No similar episodes found.

Similar Podcasts

Tales Of A Superstar DJ The Insomniac Spun seemingly out of nowhere from her complacent life in the corporate world, turned seemingly overnight from 16-Hour shift work and into the life of a literally starving artist and working musician, The Protagonist navigates her supposed rise to fame and superstardom on a journey through spiritual awakening, coming-of-age, and intimate self-realization--guided by an omnipresent force and equipped with the power of love, magic, and music. {Enter The Multiverse.} [The Festival Project] The Festival Project, Inc.™ is a multidimensional multimedia platform which encompasses exploratory and artistic social personifications and expressions on cosmic theory, spirituality, growth, health & wellness, philosophy and theoretic dynamics in entertainment such as music, design, film, television, radio, dance and festival culture, art, fashion, literature, and science. The Festival Project™ and its subsidiary Non-Profit, The Collective Complex © aims to challenge modern artistic and philosop Explicit Bitcoin Is Dead Trey Carson Welcome to Bitcoin is Dead, the ultimate Bitcoin variety show where host Trey takes you on a journey through the ever-evolving world of Bitcoin. Each episode brings new personalities, fascinating locations, and insightful conversations with politicians, educators, and innovators shaping the future of Bitcoin. Whether you're a seasoned Bitcoiner or just starting your journey, tune in for thought-provoking discussions, unique perspectives, and a deep dive into the ideas and people driving the Bitcoin revolution. Explicit Northern Sass and Southern Class Tay and Ani Come sit in on girl talk with Tay and Ani as we discuss life in Texas, girl math, food, wine and roasting each other. Explicit Never Time to Give Up Shadoe Lass A nod to the classics with a note from the future. A project meant to encompass every call I wanted to make but never went through. Seriously, it's just me, calling you. Pick up the phone? :) Hosted on Acast. See acast.com/privacy for more information. Explicit

Frequently Asked Questions

How long is this episode of Clearer Thinking with Spencer Greenberg?

This episode is 1 hour and 19 minutes long.

When was this Clearer Thinking with Spencer Greenberg episode published?

This episode was published on January 24, 2024.

What is this episode about?

Can I download this Clearer Thinking with Spencer Greenberg episode?

Yes, you can download this episode by clicking the download button on the episode player, or subscribe to the podcast in your preferred podcast app for automatic downloads.

URL copied to clipboard!