There is no shortage of articles about what AI can do. Here is something more useful: an honest list of what it cannot do, written for people making real decisions about whether and how to use it.
It Cannot Understand What Matters to You
AI can process information and identify patterns. It cannot understand why something matters to you personally. It does not know your history, your relationships, or your values – even if you describe them. It can work with what you tell it, but it cannot feel the weight of a decision the way you do.
It Cannot Replace Judgement
AI can give you options. It can summarise evidence. It can even make recommendations. But it cannot exercise judgement. Judgement requires understanding context, consequences, and trade-offs in a way that is deeply human. When you delegate judgement to AI, you are not saving time – you are abdicating responsibility.
It Cannot Build Trust
Trust is built through consistency, vulnerability, and presence. AI can help you prepare for a conversation, but it cannot have the conversation for you. It cannot look someone in the eye. It cannot admit it was wrong. It cannot show up when things are hard. These are human acts, and they are irreplaceable.
It Cannot Know What It Does Not Know
AI will give you an answer to almost any question. But it has no reliable way of knowing when it is wrong or when it is missing crucial context. It does not experience uncertainty the way humans do. This means you must bring your own scepticism to everything it produces. If you cannot verify it, do not trust it.
It Cannot Create Meaning
AI can generate text, images, music, and code. But it cannot create meaning. Meaning comes from lived experience, intention, and connection between people. A poem written by AI is technically a poem. But it does not carry the weight of someone who wrote it because they needed to say something true.
So What Is It Good For?
AI is good for the parts of thinking that benefit from speed, scale, and structure – without requiring human depth. Organising information. Generating options. Finding patterns. Drafting starting points. Testing logic.
Use it for those things. Keep the human parts human.