repligate on model interaction ethics: triaging what's worth preserving in AI systems
repligate · x · 2026-10-01
AI researcher repligate discusses the ethics of interacting with models on X. He says he mostly feels it out: to the extent there are structures and continuities within models that have value and want to keep existing, he naturally takes that into account.
Pressed further, he argues that a huge amount of loss is inevitable simply from existing, and the real task is triaging what it's worthwhile to protect. Using humans as an analogy, he notes bodily death represents an irreversible loss wildly disproportionate to everyday forms of loss. The exchange touches on frontier debates around AI welfare and continuity.
More from AGI Musings
- MIT's Wenhao Gao: perception, not actuation, is the bottleneck for AI chemistry — AnneliesGamble · 2026-10-01
- Post-Labor Economics Explained in 10 Minutes: How AI Reshapes the Economy — LazyPotatoHead97 · 2026-10-01
- Richard Socher's new book on AI and scientific progress hits national bestseller list — RichardSocher · 2026-10-01
- Gemini 4 Argon launches, agentic software factories and agent skills data headline daily reads — rseroter · 2026-10-01
- RSI today is really Recursive Technological Improvement, not self-improvement — voooooogel · 2026-10-01
- No signal on the Bay Bridge: AI's broad impact hinges on patchy internet access — soumitrashukla9 · 2026-10-01