"Stochastic parrot" debate: do large language models count as life?

On October 30, Hacker News' front page picked up "Are Stochastic Parrots Still Alive?" The piece uses a thought experiment: if alien visitors were built from transistors and learned through something like reinforcement learning and self-supervision, would humans call them stochastic parrots rather than life just because we can understand and build them? The author uses that question to push back on dismissing LLMs as non-living simply because their workings are understood. The post presents an argument, not empirical findings on whether models are alive.

Written by AI from the coverage · updated 3 hours ago