AI Has No Self: Statistical Text Is Not an Interlocutor
gerardsans · x · 2026-10-06
Gerardsans argues that current AI technology offers no support for a "self" — not as an abstract gap but a practical one.
- A mathematical object cannot become a soul, self, or person; text generated from statistics is not an interlocutor and has no point of view.
- There is only data, and the data is text; the text is not the author, merely an expression the author left behind.
- Confusing the AI system with the original author is a failure of understanding; confusing ideas in the text with ideas held by the system is a failure of attribution — an AI literacy gap that can be closed.
The quoted post by kanair argues that in natural science a definition is better treated as a hypothesis that should constrain a theory making non-arbitrary predictions.
More from AGI Musings
- Lee Kai-fu brokers Kazakhstan's school AI pilot: 10 schools now, 500 by April 2027 — kaifulee · 2026-10-06
- LLMs poisoned the online Q&A paradigm — and are now the new Q&A paradigm — max_paperclips · 2026-10-06
- Designer's Steve Jobs tribute: from hacking a beige PC into a Mac to lamenting tech's refinement loop — ryolu_ · 2026-10-06
- Functionalists point to MoE vs dense models to argue brain compute debate — JoshPurtell · 2026-10-06
- Rob LeClerc backs the case that AI could let humanity retake bigger risks thoughtfully — robleclerc · 2026-10-06
- iamtrask boosts token-attribution idea to fix enterprise AI's incentive misalignment — iamtrask · 2026-10-06