Chat GPT 5.4 solved a 60+ years unsolved erdos problems in a single shot

For years, the AI/ LLM critics had the same reasoning: LLMs don't reason and they just predict the next token Recently, it reasoned better than 50 years of mathematicians on an open erdos problems by applying a basic phd level formula Chat gpt conversation: https://chatgpt.com/share/69dd1c83-b164-8385-bf2e-8533e9baba9c Here is the problem where TAO also commented on it: https://www.erdosproblems.com/1196 Thoughts?