@Michael Loucks
I've gone the full 15 with Grammarly support on their claim that clicking 'Dismiss' makes Grammarly 'smarter'. Not even close.
Worse, they close tickets as 'resolved' when all they've done is (per their own claims) sent them to the development team. No further feedback, no release notes, just radio silence.
And I don't see how it can get better. And I can pretty much see a huge problem with AI in general, and it can be described in two words. Feedback Loop.
I take part in several forums dedicated to science and military topics. And the amount of AI being used to write the articles now is almost frightening. At least 3 "authors" I have found of articles relating to geology have now come back as 100% AI. I first realized this when I was reading an article about something in San Francisco geology, and a few months later a different article from a different author was saying almost the same thing about a city in China. I plugged them both into an AI detector, and both came back as AI.
Now we all know that AI learns through searching other things. How much of the very "learning models" they say they are using are learning from other AI? That is pretty much guaranteed that they are not learning anything, but are in a loop that will keep it stagnant and even regress.
As an IT person, I have been playing with AI for over four decades now. And I see it as no better than Eliza was in the 1970s. I've seen the video game bubble, the dot com bubble, now I believe we are in the "AI Bubble". And eventually people and companies are going to realize it is all smoke and mirrors and it will crash.
After all, if it really is "learning", why did they have to lobotomize it to stop telling people to use gasoline in recipes? That is just one example of the problem with AI. Somebody asked about cooking with gas (meaning natural gas), and it returned suggestions for cooking with gasoline as an ingredient. That got passed around, more people started to "troll" the AI, and the developers realized that no matter what they tried to do, the AI kept insisting that gasoline was fine in a recipe and you could safely eat the food it was used in.
They finally had to lobotomize AI search engines to simply ignore the question. Use any AI powered search engine like Google or Bing, they shut off the AI response because it insisted gasoline was safe to consume.