Crossposted from https://sopuli.xyz/post/50018032
The research team’s conclusion was blunt: “No tested version of ChatGPT can reliably generate appropriate responses to psychotic content.”
…
“Is a mental health AI better than a chatbot?” Torous said. “Is it better than Tetris? I think we have to prove their benefit in a rigorous way.”
Reminder -> check out the Arstechnica comments, they are always interesting and give more usually expert perspective.
Kinda silly title. The question whether something can be done, or is being tackled in real life, are two very different questions.
Last year when I read about AI psychosis, the state of the world was, there’s been some lawsuits after people killed themselves. And we learned OpenAI’s internal systems even worked(!) They flagged that account hundreds of times. It’s just that they didn’t do anything about it. So it took a turn for the worse.
Dunno if that changed since. I bet they’re not letting anyone have a look unless there’s some big lawsuit… But why argue hypotheticals from a world which doesn’t exist? First thing they need to do is do something in the first place. We probably can’t answer the question from the title until someone puts in some effort and tries.