OpenAI's ChatGPT and Codex were referring a lot to goblins, trolls, and raccoons earlier this spring. And that happened inadvertently when the AI lab prompted those models to have a "nerdy" personality, says Janelle Shane, writer of the blog "AI Weirdness."
The problem of AI alignment has become a hot topic. It’s raised questions, like how do we get AI to do what we want, and not what we don't want? For example, hacking another company. Or, in a less criminal vein, how do we get AI to stop talking about goblins?
After mounting user reports of this behavior earlier this year, OpenAI had to give its models — ChatGPT and Codex — explicit instructions to cut it out. It's the topic of today's “Uncanny AI,” those moments when it's clear AI doesn't think like us.
Janelle Shane, author of the blog "AI Weirdness,” says goblin mode emerged from a nerdy personality customization. “Marketplace Tech” host Meghan McCarty Carino asked her why.
“OpenAI Really Wants Codex to Shut Up About Goblins” from Wired
“AIs have ‘personalities’ – here’s how they affect you more deeply than you may realize” from The Conversation
Apple
Spotify
Amazon
RSS