I wonder to what extent the future is unknown for ChatGTP? In general, all events range from "can't happen" to "will definitely happen". Obviously if you ask "will the sun rise tomorrow" then the answer will be "Yes" and not "I am a language model so I don't know". Probably with sporting events it is possible to get some thoughts from ChatGTP. But it is necessary to correctly formulate questions and not ask what is really impossible to predict.
Maybe it's possible to have some thoughts when making questions regards each factor in particular which can affect game's result. Like, who is injured in team A? What is the weather forecast for tomorrow in X area? What is the currently winning and loss streak of team B?
Actually, I still doubt the AI has conditions to answer this kind of question accurately right now, but in some updates from now on I guess it will be pretty possible.
Anyway, be careful with ChatGPT. It has already given me many wrong replies, misunderstanding people's names and what they did.
I think more general questions can be asked. Of course, not such as "what will be the score in the game City - Newcastle", but for example "what is the probability of a draw in the game City - Newcastle". I see no reason why ChatGTP cannot, by checking publicly available statistics, give at least a range of their assumptions. And in the future, I am sure such tools will give more and more accurate answers.
The language model only covers data up until sometime in 2021. So data that would go further than said year would be beyond ChatGPT's means. There's also the case of stuff that it has no power over knowing because data about it is really nonexistent, like asking it what is the quotient of 0/0. I reckon and I tried this quite a few times already, that ChatGPT could still provide game stats for people who are really looking to use it as a form of prediction tool to know game outcomes before the match begins, but it is also worth noting that these data are fairly outdate and may not really mean shit in the grand scale of things, but oh well, If you're doing it for the kicks might as well go for it, but as long as openai is stuck with ChatGPt and its outdated sources, you can't really use it for game predictions.
Thanks for the details, I didn't know that. But these are, let's say, "temporary" problems. The tool will undoubtedly improve and in a year or two it will search for (and, most importantly, systematize) information faster than people.