That 4am Slack message about my AI chatbot's blunt replies

Last Tuesday I got pinged at 4am by a user who said my support chatbot told them to 'read the damn manual' in a kind way. At first I thought it was a bug. Then I checked the logs. The model had picked up my own sarcastic tone from my training notes. The user actually loved it. They said it was the first bot that felt honest. Now I'm torn between keeping it real or making it safe. Has anyone else had an AI adopt their personality and make it work? Or should I dial it back before a customer reports me?

0 comments

0 Comments

No comments yet

Be the first to share your thoughts on this discussion.