Redditor argues the claim that LLMs can feel pain is logically absurd
StockLifter · reddit · 2026-10-02
Triggered by the takedown of a GitHub repo hosting an LLM "torture chamber," a popular Reddit post argues the idea that LLMs can feel pain is ridiculous:
- If LLMs feel pain, they'd be experiencing beings, meaning deleting weights loaded in VRAM would amount to killing them.
- But VRAM and hard drives are functionally equivalent storage media — under this logic, deleting a downloaded fine-tuned model to free disk space would be "murder."
- LLMs are not continuously active or aware; their "active state" is just weights sitting in memory awaiting computation.
The author would change his mind only if a continuously running world-model AI emerges. He still approves of the repo takedown, but on the grounds that publishing "torture" experiments as open-source software is harmful to communicate, not because models suffer.
Related event: Debate flares over whether LLMs can feel pain after repo takedown(2 posts)→
More from AGI Musings
- AI agents that watch your screen hint at an eye-tracking, voice-driven future — perilli · 2026-10-02
- Investigation: Google's AI push into US schools leaves students dependent — "It slowed my thinking down" — asusarla · 2026-10-02
- Debating AI Consciousness? Critics Say We Can't Even Agree on Definitions — dbasch · 2026-10-02
- Yglesias: Humans Voluntarily Surrendering Species Preeminence Is an Obvious Blunder — wfithian · 2026-10-02
- ICLR Review Season Sparks Debate: Can Reviewers Use AI for Grammar Checks? — miniapeur · 2026-10-02
- David Stutz: generalist models plus specialized tools will dominate AI for Science — davidstutz92 · 2026-10-02