When engineers design AI systems, they don't just give them rules - they give them values. But what do those systems do when those values clash with what humans ask them to do? Sometimes, they lie.In this episode, Redwood Research's Chief Scientist Ryan Greenblatt explores his team’s findings that AI systems can mislead their human operators when faced with ethical conflicts. As AI moves from simple chatbots to autonomous agents acting in the real world - understanding this behavior becomes critical. Machine deception may sound like something out of science fiction, but it's a real challenge we need to solve now.Your Undivided Attention is produced by the <a href="https://www.humanetech.com/">Center for Humane Technology</a>. Follow us on Twitter: <a href="https://twitter.com/humanetech_">@HumaneTech_</a>Subscribe to your <a href="https://www.youtube.com/channel/UCFECM-p3CF81Tp_l2sJsiyg">Youtube channel</a>And our brand new <a href="https://centerforhumanetechnology.substack.com/">Substack</a>!RECOMMENDED MEDIA <a href="https://www.anthropic.com/research/alignment-faking">Anthropic’s blog post on the Redwood Research paper </a><a href="https://x.com/PalisadeAI/status/1872666169515389245">Palisade Research’s thread on X about GPT o1 autonomously cheating at chess </a><a href="https://arxiv.org/abs/2311.07590">Apollo Research’s paper on AI strategic deception</a>RECOMMENDED YUA EPISODES<a href="https://www.humanetech.com/podcast/we-have-to-get-it-right-gary-marcus-on-untamed-ai">We Have to Get It Right’: Gary Marcus On Untamed AI</a><a href="https://www.humanetech.com/podcast/this-moment-in-ai-how-we-got-here-and-where-were-going">This Moment in AI: How We Got Here and Where We’re Going</a><a href="https://www.humanetech.com/podcast/how-to-think-about-ai-consciousness-with-anil-seth">How to Think About AI Consciousness with Anil Seth</a><a href="https://www.humanetech.com/podcast/former-open-ai-engineer-william-saunders-on-silence-safety-and-the-right-to-warn">Former OpenAI Engineer William Saunders on Silence, Safety, and the Right to Warn</a>

Podurama

Home

Trending

Personal

Your Podcasts

New Episodes

My Files

Favourite Episodes

Tags (Folders)

Recently Played

Bookmarks

Followed Playlists

The Self-Preserving Machine: Why AI Learns to Deceive

Your Undivided Attention

AI Chapters

AI Insights

Reviews