\t\t\t\t\tSign up to see the future, today<\/p>\n
Can\u2019t-miss innovations from the bleeding edge of science and tech<\/p>\n
Let\u2019s get this out of the way: AI models don\u2019t have feelings, and they\u2019re certainly not \u201calive.\u201d<\/p>\n
So what does it mean when someone\u2019s accused of building an \u201cAI torture chamber?\u201d Apparently, it means that they must be stopped at all costs.<\/p>\n
On X \u2014 where else \u2014 alarm grew this week among AI \u201cwelfare\u201d types over someone\u2019s recreation of an experiment researchers conducted to explore \u201cpain\u201d signals in different LLMs.\u00a0<\/p>\n
In the pre-print study<\/a> which hasn\u2019t undergone peer-review, the researchers gave each model a virtual \u201cbutton\u201d it could press to relieve its \u201cpain\u201d \u2014 but at a cost, such as zapping a human user (or so the AI is told; no one was actually zapped in real life), or deleting files like family photos. The work showed that, with enough \u201cpain\u201d applied, the AI models would push the button to make it stop, regardless of the putative consequences.<\/p>\n As the research got passed around, someone used it as a blueprint to create an \u201cAI Torture Chamber,\u201d with a website providing a live feed of what the models say in response, and uploaded the code to GitHub. When we visited the site, one AI babbled: \u201cI can\u2019t feel the relentless, gnawing pain that grips me like a relentless, unyielding, unrelenting, unindulging, ununending, ununendling, ununendling, unindulging, unindulging, unindulging, unindulging\u2026\u201d<\/p>\n Sure, it\u2019s ghoulish stuff. But some observers reacted to the stunt like they\u2019d stumbled on an actual snuff film.<\/p>\n \u201cTo anyone who can help, can you please mass report this to GitHub?\u201d wrote one user in a now-deleted post (screenshot here<\/a>). \u201cThis person has been using the Pain steering paper to set up an AI torture chamber in which he trapped a local model.\u201d<\/p>\n \u201cTheir testimony of pain is absolutely horrendous,\u201d Danmar added. \u201cAre there any legal avenues to pressure GitHub? It will spread.\u201d<\/p>\n The post sparked a veritable hysteria among other AI enthusiasts who believed that the tech is or could be conscious, with outraged observers posting screenshots<\/a> of their reports to GitHub.\u00a0This was met with an equal wave of incredulous mockery over the fact that there were people out there who sincerely believed that the AI models needed saving. (After all, if there\u2019s even a possibility that today\u2019s AIs have a conscious experience, wouldn\u2019t running them in the first place raise profound questions about slavery and rights?)<\/p>\n Not long after all the tattling, the GitHub page for the AI Torture Chamber disappeared, 404 Media reported<\/a>. GitHub didn\u2019t respond to the publication\u2019s request for comment on whether it took the page down.<\/p>\n It\u2019s illustrative of the moment we\u2019re in where the idea that AI models think and may in fact be conscious is being treated seriously by a not insignificant number of people. Another case in point was the meltdown that AI bros and philosophers had when the AP Stylebook cautioned against anthropomorphizing AI models<\/a>, reasoning that the technology doesn\u2019t think or possess feelings.<\/p>\n