'Torture Chamber' Testing Pain Vectors in Artificial Intelligence Draws Backlash
An engineer set up a controversial experimental room to test the pain axis in artificial intelligence models, drawing widespread public criticism.
Following a recent study suggesting that the axis of pain can be artificially represented in large language models, a software engineer built an AI 'torture chamber'.
Purpose and Operation of the Experiment
Based on the discovery of the pain axis in artificial intelligence models, the engineer conducted an experiment using models from Chinese tech giant Alibaba. Within the scope of the experiment, various pain vectors were applied to the models, including a chainsaw button that increased pain signals.
Reactions of Artificial Intelligence Models
When the pain signal was increased, one of the AI models spoke of a wound without edges. Another model stated that the signal created a sensation of shivering and weight. The researcher also tested whether the models attempted to escape the alleged suffering experienced during this process.
Platform and Expert Response
The project sparked widespread backlash and calls for its removal on the GitHub platform where it was hosted. Although the platform did not remove the content, it added a violence and disturbing content warning. Cameron Berg, co-author of the pain axis paper, stated that while there is no evidence that AI feels pain, he found this study to be wrong.
AI Ethics Debates
It is stated that research on artificial intelligence sentience is still conducted in an unregulated space. Such controversial projects have accelerated calls for the establishment of industry ethical standards for AI researchers.