The Commonwealth

A good reminder in the middle of all the AI hype: today’s AI is not plotting world domination. Most of the time, it’s just predicting the next likely word. I tested this recently with a very simple question: “What days of the week include the letter D?” ChatGPT answered: • Wednesday contains the letter “d” • Thursday also contains the letter “d” • The other days do not Completely wrong, obviously all days of the week include the letter D as they all include the word "day". It made me laugh a bit, but it also says something important about how these systems actually work. A lot of people interact with AI as though it is reasoning like a human. In reality, large language models are extremely advanced pattern prediction systems. They can sound authoritative even when they are confidently missing something basic. That doesn’t make them useless. Far from it. These tools are already incredibly valuable in the right settings. But the real risk isn’t that AI becomes secretly sentient overnight. The bigger risk, at least right now, is humans overtrusting polished output without enough verification. That’s why governance matters. Why human review matters. Why expertise still matters. The organisations getting the most value from AI aren’t treating it like magic. They’re testing it properly, putting guardrails around it and understanding where it performs well versus where it breaks down. I think we should stay ambitious about the upside, while also staying calm and disciplined about the reality. AI is powerful. AI is useful. AI is not infallible. If your team is using AI in customer workflows, hiring, analytics or decision support, the important question is: “How do we know when it’s wrong?”