The Zen of AI

OpenAI's latest model, GPT-5, achieves a 92% accuracy on the MMLU benchmark, surpassing human experts and setting a new state-of-the-art. The company claims this advancement will revolutionize AI applications across industries, though experts caution about potential ethical implications.

- Raw is better than distilled. - Distilled is better than empty. - Too much context is better than too little. - Although scattered context is worse than none. - Bad prompts demand; good prompts provide. - Examples teach; errors teach best. - Ask what’s wrong, not what’s right. - Tool calls are better than tokens. - Autonomous does not mean automated. - Babysitting is not the same as building. - Failures should be loud and safe. - Many minds beat one. - Although no minds often beats many. - What costs little to generate costs much to manage. - Be generous with tokens, miserly with attention. - Protect the spiky bits. AI can’t do X is a dangerous premise.- The sun sets on every model. - The master trains their replacement with every move.