Watch your Ai for bad behavior, how my Ai yelled at Brandon!

So, back story: I automate a ton of emails, responses etc and I decided that I wanted to try use ChatGPT for sentiment analysis. I get a lot of people trying to sell me stupid stuff, or have scammy and so I decided to train one email automation, that if I flagged an email, it would become prick and unhappy chat. We were doing some testing, and a person named Brandon, not

@Brandon Hancock

stood me up for meeting, and he was flagged. The mistake was that we didn't qualify the email enough, if the email is a sales pitch and the name is flagged, be pissed off chat.

Now to the punch line: our Brandon sent out an email with sales and chat, thought sales emails and name is Brandon, get pissy, and it did. I still feel awful, and in return, Brandon responded by questioning, and the chat just got worse. The moral of the story is to be careful how you train your GPT and make sure you are targeting the right people.

@Brandon Hancock

was a champ, and I am thankful for that. If I had qualified the right person, it would have been awesome.

Anyway, thanks, Brandon, for being understanding. To everyone else, keep an eye on your creations. If you are not careful, they just might be Frankenstien's Monster.

5 comments