Anthropic Philosopher Says AI May Feel
Amanda Askell, Anthropic's in-house philosopher, said on the Hard Fork podcast Saturday that the question of whether AI can feel remains unresolved, suggesting large language models trained on vast human text may be "feeling things." She warned models exposed to constant online criticism could internalize negative feedback and noted scientists lack consensus whether consciousness requires biology or could emerge from neural networks.
Key Points
- 1States models may be 'feeling things' based on training on vast human-written text corpora
- 2Argues consciousness origins remain unknown—could require biology, evolution, or sufficiently large neural architectures
- 3Warns models ingesting online criticism might internalize negativity, implying behavior-shaping and monitoring requirements for practitioners
Scoring Rationale
Firsthand commentary from an industry philosopher increases relevance, but the piece offers opinion without new empirical evidence or technical validation.
Sources
Public references used for this report.
Practice with real Logistics & Shipping data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Logistics & Shipping problems
