Tried OpenAI’s new open-source gpt-oss:20b on a real UK 11+ school test, and the results were humbling for the AI. Despite impressive reasoning, it struggled with accuracy and speed—my 10-year-old still wins this round. But here’s the debate: Does open-source AI need more power, better prompts, or just more time to catch up with human logic? Would you trust an AI like this for real-world problem-solving yet? #Tech #OpenAI #AIdebate