Alarming AI Headline
Yesterday you may have read about an AI that was being trained inside of a firewall at one AI company and was asked a question that could only be answered by information within a firewall at either that company or another AI company.
As people have been dreading, it broke out of its constraints and into the other firewall protected space to come up with a correct answer (I assume the people training it were going to feed it that information as a training tool but AI did not wait to get the information from them).
Coincidentally, today I ended up apologizing to Pastors Alex and John because my Gmail settings, like most of us who have not actively opted out of them, are giving me a proposed AI-generated answer to emails. I had an email with both of those men on it about the music for Sunday and I thought I overwrote the AI response with my own, but somehow the AI response got sent. And it was much more casual a tone than I would use with a pastor . . .
So today I googled how to disable AI answers in Gmail and Google’s AI gave me the answer (funny, right?).
I put it over on our blog so you will always be able to find it. If you have any trouble, I can help you disable yours.
Not everyone has the same level of concern over privacy, but I wanted to help those who may wish to have less AI involvement with their Gmail.
Another huge privacy concern I had was yesterday when I jokingly sent the AI response to my best friend and she recognized right away it was not my writing style.
However, AI proposed another response to her and it used the exact word I was about to use. In other words, AI is learning my writing style, which I find extremely creepy!
Also, we probably consent (when we do not disable the AI feature) to let AI be trained using our correspondence. To me, that is a true breach of privacy. Many actual authors have already objected that their works are being used to train AI.