AI doing philosophy = AI generating hands? — LessWrong

I've been playing around with Stable Diffusion recently, and an analogy occurred to me between today's AI's notoriously bad generation of hands and f…

x AI doing philosophy = AI generating hands? — LessWrong AI Capabilities Intellectual Progress (Society-Level) Meta-Philosophy Philosophy AI Frontpage 51 AI doing philosophy = AI generating hands? by Wei Dai 15th Jan 2024 AI Alignment Forum 3 min read 26 51 Ω 22 I've been playing around with Stable Diffusion recently, and an analogy occurred to me between today's AI's notoriously bad generation of hands and future AI's potentially bad reasoning about philosophy. In case you aren't already familiar, currently available image generation AIs are very prone to outputting bad hands, e.g., ones with

related reading

• Current AIs seem pretty misaligned to me — LessWronglesswrong.com
• My picture of the present in AI — LessWronglesswrong.com
• Deep Deceptiveness — LessWronglesswrong.com
• The Argument from Philosophical Difficulty — LessWronglesswrong.com
• What failure looks like — AI Alignment Forumalignmentforum.org
• How do we (more) safely defer to AIs? — LessWronglesswrong.com
• Why You Should Almost Never Use AI to Write Anything Substantiveerichgrunewald.com
• AI Safety for Fleshy Humans: a whirlwind touraisafety.dance
• The Machines Lack Honour — LessWronglesswrong.com
• The Universe from an Intentional Stancecasparoesterheld.com
• Many AI Debates Are Way Too Philosophical and Theoreticalbenthams.substack.com
• Wei Dai's Shortform — LessWronglesswrong.com