AI: The not-so-good parts Posted by luu on Wednesday January 17, 2024 @09:04PM from the xeiaso.net dept. https://xeiaso.net/talks/2024/prepare-unforeseen-consequences/ ( 1 comment | 2 points | on HN )
There are problems with guardrails that are prominent across gpt - chatgpt and gpt other versions. even 4 can't be trusted with predictable outcomes.