|
Artificial “Intelligence” is not very intelligent
I’ve been really surprised at how dumb AI actually is. Both ChatGPT and Grok. When I read how much people are relying on its “smarts,” it’s really perplexing. I read about lawyers using it, doctors etc. Frightening.
For example:
I asked it “has Bertone ever designed a Ferrari?” It said no, Bertone never did. I pointed out that Bertone designed the mid 70s 308 GT4, and it said “oh, yes, you’re right, sorry about that.” Huh? It couldn’t accurately research a super basic question that has been discussed on the Internet for decades?
I asked it to give me the plot of a movie that I had just watched on Prime. Not a big blockbuster type movie. It gave me a whole long detailed and authoritative plot description. Except it had almost no relationship to what actually happened in the movie. It was 80% made up stuff!
I asked “why did you make all that up??” It said it couldn’t find much info about the movie so it just made up things that were “likely” to have happened. I asked why it would not just say it didn’t know or couldn’t find any info. It said “I’m programmed to make up info I don’t know because our research indicates that enhances the customer experience, people are not satisfied with ‘I don’t know’ as an answer.” Wow.
It’s also really bad with current events. I asked what time a baseball game was on TV, and it gave an obviously wrong answer. I told it it was wrong, and it rechecked and insisted it was correct, twice. The only way it would admit it was wrong was when I gave it the correct answer.
Seems kind of useless to me. It certainly can’t be relied on in any way for accuracy.
|