AI Models from Google, OpenAI, Anthropic Solve 0% of ‘Hard’ Coding Problems

Despite claims of AI models surpassing elite humans, 'a significant gap still remains, particularly in areas demanding novel insights.'

4 comments

But they have "reasoning" models!
- I love asking those to explain their reasoning.
  
  Then explain it again.
  
  And again.
  
  Ten times in a row.
  
  Then count the different "reasonings" and especially the mutually incompatible ones.
no cs grad who does their homework with chatgpt can either
Well this is just one objective test from a bunch of independent universities.

It's not the same as a never ending stream of self-tests by people profiting from the grift. That "research" is worth hyping all over the place! It's super promising and worth bazillions in VC funding. \s