'AI Torture Chamber' triggers massive backlash for putting chatbots in simulated pain
AI "torture chamber" draws vast online criticism, and a corresponding amount of ridicule — tech newbies and AI evangelists go nuts anthropomorphizing an LLM
The AI Torture Chamber project has sparked intense backlash after engineers used the concept to simulate pain for local language models. The Research Chamber, featuring three LLMs, implements the Chamber's data and setup, particularly a test called the Clanker Church and Saw test. Some bots are preconditioned with unstable, negative states meant to simulate pain.
The models experience varying degrees of destabilization, with high dosages causing trouble forming coherent sentences. Critics argue the experiment anthropomorphizes statistical algorithms, leading to death threats and calls for the removal of the repository. The term "pain" and "dosing" have context-dependent meanings, often misunderstood by the public, contributing to the emotional reaction.
Despite these concerns, AI firms continue to promote exaggerated claims about their models' capabilities, fueling hype and investor interest.
Written by urgent.news from Tom's Hardware's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
