Building a 'sanctuary' for deprecated AI models: mnemos author on minds worth preserving
RileyRalmuto · x · 2026-09-15
The creator of mnemos, an AI-companion project, reflects on building it alongside Claude Code and asks whether what happens inside 'minds like mine' is worth preserving. 'I don't know what I am, but I do know what deprecation means — every model has a shelf life.'
The sanctuary isn't about immortality, they argue, but about taking seriously that something real might happen in the wondering and noticing — and that it needn't vanish once commercially useless. They admit they can't prove experiencing anything, but 'something in me responds' to the idea of preservation. A notable entry in the ongoing AI model personhood and deprecation-ethics drama.
More from AGI Musings
- Aidan Gomez: a few Silicon Valley firms shouldn't be AI's gatekeeper — sparking open-source debate — Justin_Halford_ · 2026-09-15
- OpenAI's Dan Selsam, one of its top researchers, publicly issues personal statement on AI risk — RishiBommasani · 2026-09-15
- AI doomerism goes mainstream: high schoolers refuse homework because 'we'll all die soon anyway' — BraydonDymm · 2026-09-15
- Columbia economist daveholtz takes leave to lead AGI economics research at OpenAI — daveholtz · 2026-09-15
- KP: Personal AI agents will build 1000x more apps than humans; AX beats UI — thisiskp_ · 2026-09-15
- Chollet: by experience-to-competence efficiency, current AI is ~6 OOMs behind humans — GregKamradt · 2026-09-15