The level that some AIs have achieved is mind boggling

I have been fortunate enough to work on a few specialized AI models designed specifically for logic and they blew my expectations away. These models take a significant amount of time in generating a response, but the correctness in logic and conciseness of the final output was very good. I needed to think of very complex problems (while being reasonable in scope for the AI) before the AI actually started to mistakes.