ChatGPT says we’re #1
- Thread starter CatManTrue
- Start date
Rutgers as the meanest? I guess they should wait for the next version of Chat GPT.
That was one of the rankings which surprised me. OSU and Nebraska’s also did. ChatGPT may have started hallucinating after they got the first one right.
This is the problem with ChatGPT. It can gather data, but it can't make certain distinctions. In this case, ChatGPT probably saw that if you're a visiting fan to an NU football game, the majority of the crowd is also for the visiting team, so they interpreted that to be the friendliest scenario.
It’s interesting to think what data they used to arrive at that ranking. I doubt they had any legitimate sources for breakdown of game attendance.
They certainly used Twitter, Reddit, and other “open” sources as much as they could. Our fanbase is sparse and also nicer than most if not all P5 fanbases so those would have helped our case.
I bet they also used the public boards like these. In which case… even with the Rant Board, our fanbase is still way kinder than others.
The chant..."“That's alright, that's okay, you're gonna work for us someday!”...almost cost us 1st Place.
DeepNerd alert here.
I run a company that is built on the same type of Large Language Models that made ChatGPT possible. Unlike ChatGPT, we apply them as an earpiece (eg analytical tool) rather than a mouthpiece (chat or writing tool). Here's an expert view on the above ranking of fan friendliness.
Because ChatGPT is built to talk and has to respond in milliseconds (or write the next sentence in sequence in milliseconds) it can not conceivably check all the relevant facts on the interweb before it provides a confident answer. It has no choice but to make stuff up, because of how it's built.
(E.g. Rutgers, OSU, Nebraska per the above.)
That means using chat tools to analyze anything is dumb. By contrast, using Large Language models as an earpiece, you get things that are impossible to get from a chat tool. (We've been working with Twitter for a few years to provide predictive understanding of narrative trends/patterns within their platform globally)
Net - as an AI practitioner I'd argue that Artificial Intelligence is often intelligent. But its dumb as hell if you build it to talk, then ask it to analyse. Its like firing your best analysts and BI people, then replacing them with the corporate coms people.
Standing offer - If anyone has a bunch of unstructured natural-language data (all comments for the past 5 years on the basketball board for example) I can dump it into our tool and share the analysis with you all.
I run a company that is built on the same type of Large Language Models that made ChatGPT possible. Unlike ChatGPT, we apply them as an earpiece (eg analytical tool) rather than a mouthpiece (chat or writing tool). Here's an expert view on the above ranking of fan friendliness.
Because ChatGPT is built to talk and has to respond in milliseconds (or write the next sentence in sequence in milliseconds) it can not conceivably check all the relevant facts on the interweb before it provides a confident answer. It has no choice but to make stuff up, because of how it's built.
(E.g. Rutgers, OSU, Nebraska per the above.)
That means using chat tools to analyze anything is dumb. By contrast, using Large Language models as an earpiece, you get things that are impossible to get from a chat tool. (We've been working with Twitter for a few years to provide predictive understanding of narrative trends/patterns within their platform globally)
Net - as an AI practitioner I'd argue that Artificial Intelligence is often intelligent. But its dumb as hell if you build it to talk, then ask it to analyse. Its like firing your best analysts and BI people, then replacing them with the corporate coms people.
Standing offer - If anyone has a bunch of unstructured natural-language data (all comments for the past 5 years on the basketball board for example) I can dump it into our tool and share the analysis with you all.
I already did this. It said I was the wisest poster on the board and that you should all listen to me!
