AI gone wild: let’s discuss times AI got it really wrong

In response to last week’s letter about a manager who didn’t want people using AI for note-taking at meetings, some readers shared particularly ridiculous firsthand examples of AI getting it wrong. For example:

• Ours once transcribed a side conversation about my water bottle we had while waiting for someone to arrive, and then assumed the entire meeting, which was actually about software design, was about the water bottle.

• I would like to shout-out the AI transcription tool at my old job that took notes at a meeting evaluating applicants for a job…and then automatically emailed said notes to the entire company AND to the candidates under discussion.

• I once read an AI transcript – made using an approved tool at my then-workplace – that seemed like it needed to go to HR ASAP. All this stuff about kissing and crying for mercy and whatnot. But I also listened to the audio. The problem was, the AI was transcribing in English – and the people were speaking in French. It was just making whatever it could out of sounds that had nothing to do with the language it expected. *facepalm* (“Mercy” was someone thanking someone else for some info…)

• My city council was recently busted talking trash about a resident all because there was a substitute clerk taking minutes in a closed session. Since the clerk was new, they decided to use AI to help take notes. No one would’ve known what was said had it not been discovered in a random FOIA.

• One recently recorded an informal meeting in such a way that it made it sound like one colleague said another’s dress sense was trash. It was entertaining, but there’s no way I’d trust it as a replacement for meeting minutes.

Let’s discuss times you’ve seen AI get it really wrong. Please share in the comments!

The post AI gone wild: let’s discuss times AI got it really wrong appeared first on Ask a Manager.