Wrestling Intelligence

[Wrestleknownothing]

Sorry, this one is not about that kind of wrestling intelligence.

Instead this is my screed against using artificial intelligence to answer wrestling questions. Really, though wrestling data is just my entry to the topic of how AI is currently designed to provide us with very bad answers that are very convincingly stated.

Danger Will Robinson. It is a bit of a long one. Proceed at your own risk.

How Much I, How Much A? | Wrestling.Guru

So I have been testing Anthropic's Claude AI for work. Like a good prosecutor I started by asking questions I already knew the answer to in order to es ...

wrestling.guru

 

[petethegreat]

Previous to the ncaa’s this year I asked grok to list the favorites for this year. Most of the info was correct except for 197. Where it listed rocky elam as the favorite. I asked it to do further analysis on 197 where it finally listed Josh bar as the favorite. I then asked it “How did you miss Barr in your previous analysis?”. It came back with it just missed him. Not very good for an AI.

I then asked: How can I trust you when you miss a clear favorite like that. it came back with a lot of data and said it would try to do better. I asked it what it could do to correct the issue and it gave some details like double checking and searching better but it just seemed like it was trying to list things but no changes would be really be made cause it takes money and time.

Most interactions I’ve had with AI has been unimpressive. A lot of mistakes and some outright lying or making things up. ive called it on the lying but it just says sorry and corrects it self. It seems it needs alot of improvement to be trustworthy We are a long away from AGI (Artificial General Intelligence).

 

[SleepyLion]

Previous to the ncaa’s this year I asked grok to list the favorites for this year. Most of the info was correct except for 197. Where it listed rocky elam as the favorite. I asked it to do further analysis on 197 where it finally listed Josh bar as the favorite. I then asked it “How did you miss Barr in your previous analysis?”. It came back with it just missed him. Not very good for an AI.

I then asked: How can I trust you when you miss a clear favorite like that. it came back with a lot of data and said it would try to do better. I asked it what it could do to correct the issue and it gave some details like double checking and searching better but it just seemed like it was trying to list things but no changes would be really be made cause it takes money and time.

Most interactions I’ve had with AI has been unimpressive. A lot of mistakes and some outright lying or making things up. ive called it on the lying but it just says sorry and corrects it self. It seems it needs alot of improvement to be trustworthy We are a long away from AGI (Artificial General Intelligence).

I think we are a long way from Natural General Intelligence.