It just feels too good to be true.
I’m currently using it for formatting technical texts and it’s amazing. It doesn’t generate them properly. But if I give it the bulk of the info it makes it pretty af.
Also just talking and asking for advice in the most random kinds of issues. It gives seriously good advice. But it makes me worry about whether I’m volunteering my personal problems and innermost thoughts to a company that will misuse that.
Are these concerns valid?
That’s correct, its whole experience is limited to a ~2000 word text prompt (that includes your questions, as well as previous answers). Everything else is a static model with a bit of randomness sprinkled in so it doesn’t just repeat. It doesn’t learn. It doesn’t have long term memory. Every new conversation starts from scratch.
User data might be used to fine tune future models, but it has no relevance for the current one.
This is just wrong, but despite being frequently parroted. It obviously understands a lot. Having a little bit of conversation with it should make it very clear. You can’t generate language without understanding the meaning, people have tried before and never got very far. The only problem it has is that its understanding is only of language, it doesn’t know how language relates to other sensory inputs (GPT-4 has a bit of image stuff build in, but it’s all still a work in progress). So don’t ask it to draw pictures or graphs, the results won’t be any good.
That said, it’s surprising how much knowledge it can extract just from text alone.