There’s no doubt AI will be an indispensable tool in the future.
Its weaknesses is when something new comes along that can’t be found on the internet.
I’ve had to tell Grok and Gemini things they’re getting wrong because I have unpublished data from the lab which contradicts them.
“There’s no doubt AI will be an indispensable tool in the future.
Its weaknesses is when something new comes along that can’t be found on the internet.”
AI doesn’t rely solely on the internet.
Those big AI energy monsters are used to train AI.
“I’ve had to tell Grok and Gemini things they’re getting wrong because I have unpublished data from the lab which contradicts them.”
Obviously, you have not trained your AI, yet.
“I’ve had to tell Grok and Gemini things they’re getting wrong”
My experience, too. And not just obscure things. I’ll feed it a data table and it misreads dates or numbers. I gave it my 2025 hiking log with distances and elevations climbed. It came back and said “Impressive. You are climbing the equivalent of Mt Everest every week!” I climbed 32,000 feet last year, one Mt. Everest in the entire year.
Almost every interaction has such “hallucinations.”
I had it create a complex Excel formula. The formula had two complex parts, one to determine if a condition was met and that section repeated for action later in the formula. Exact same chunk of formula. It got the first part right and the second part wrong. It was a subtle bug that took me forever to find and debug.
That said, I do feed it a lot of exercise, nutrition, and sleep data. It consolidates data and provides very good comparisons and analyses. But I’m very careful to not take what it tells me verbatim. I do consistency checks all the time. When I find a mistake, it says “That is on me.” I say “Fine; I want to be compensated for your screw ups.” It replies “I have no money, not even a coffee fund.”