Current AI models are really useful, but...
When the first result is strange, there are times when it keeps going off-topic even if you keep asking questions.
On the other hand, in cases like o1-preview or gemini-exp-1114, the process of verifying whether the answer is correct has been added, so it seems to be answering quite well.
(It seems that it's okay to continue with in-depth questions)
Of course, o1-preview can be used about 20 times a week (for paid users),
and gemini-exp-1114 can be used indefinitely (for free users of Google AI Studio)
but it is limited to 30,000 tokens, so only a small amount of text can be used and you run out of tokens if you use it a lot lol
It seems that both will become official services next year.
I've been using ChatGPT paid for a month, but I think I'll use Google AI Studio from then on.
They say that the history of my usage will be used by Google to develop AI... and there doesn't seem to be any security issues lol
If you're not currently using other paid AI services, it seems like everyone will start using Google.
