re: "AI-powered" has become a red flag
I fully agree with the premise of the post. however, I want to point out something else here.
what data do you think AI models are trained on?
the question of how IP and copyright apply to the training data is still here, kind of unsolved, but LLMs like GPT are trained on public data from the Internet, like articles, books and comments.
now think about it when a big tech messenger app announces its new GPT-powered chatbot. do you really think it is just GPT with a custom prompt and they missed on the opportunity to feed all user messages into it?
and when the messenger is Discord, and the bot is talking racial slurs, you know what happened.
but the main concern is that training data is being collected from users, and the cryptic privacy policies that the big tech adopted over the years don't help with this at all. and I promise you, they will change their privacy policies to allow for more data selling, and they'll be doing that until there is nothing more to sell, which will never happen.
we can talk Google. oh hell can we.
but it all comes down to 4 simple words: