Heavens no. I wouldn’t trust any LLM right now for factual information. Its value to me is a creative tool to in this case give you a head start on how to approach the problem. Your example of how something as simple as double counting could make the result off by an enormous factor is a perfect warning for how as well as they can bullshit it, there isn’t really as much critical “thinking” happening in that black box as some people would like to believe.
Now if we see tighter integration with Wolfram Alpha on the other hand…
Did they double count games? Since each team plays another team.
By the way never ever ever trust Bing citations. Machine learning engineers refer to them using the technical term “dog shit”.
Heavens no. I wouldn’t trust any LLM right now for factual information. Its value to me is a creative tool to in this case give you a head start on how to approach the problem. Your example of how something as simple as double counting could make the result off by an enormous factor is a perfect warning for how as well as they can bullshit it, there isn’t really as much critical “thinking” happening in that black box as some people would like to believe.
Now if we see tighter integration with Wolfram Alpha on the other hand…