AI Safety Seems Hard to Measure

Four analogies for why "We don't see any misbehavior by this AI" isn't enough.

Click lower right to download or find on Apple Podcasts, Spotify, Stitcher, etc. In previous pieces, I argued that there's a real and large risk of AI systems' developing dangerous goals of their own and defeating all of humanity - at least in the absence of specific efforts to prevent this from happening. A young, growing field of AI safety research tries to reduce this risk, by finding ways to ensure that AI systems behave as intended (rather than forming ambitious aims of their own and deceiving and manipulating humans as needed to accomplish them). Maybe we'll succeed in reducing the risk,

saved by

• Aleksi Maunu
• Lydia Nottingham

related reading

• The case for ensuring that powerful AIs are controlled — LessWronglesswrong.com
• The Case Against AI Control Research — LessWronglesswrong.com
• What failure looks like — LessWronglesswrong.com
• How do we (more) safely defer to AIs? — LessWronglesswrong.com
• What failure looks like — AI Alignment Forumalignmentforum.org
• Core views on AI safety: When, why, what, and how \ Anthropicanthropic.com
• Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover — LessWronglesswrong.com
• Recommendations for Technical AI Safety Research Directionsalignment.anthropic.com
• AI Safety for Fleshy Humans: a whirlwind touraisafety.dance
• Racing through a minefield: the AI deployment problemcold-takes.com
• The case for ensuring that powerful AIs are controlledblog.redwoodresearch.org
• Core views on AI safety: When, why, what, and how \ Anthropicanthropic.com