AI ethics is not just a problem for engineers

When people talk about the ethics of artificial intelligence, they often picture a small group of philosophers and engineers writing dense academic papers. That image is misleading. Most ethical decisions about AI are not made by ethicists at all — they are made every day by product managers, designers, salespeople and ordinary developers, often without realising it. Take the example of a model that recommends candidates for a job opening. The team building this system has to decide what counts as a 'good' candidate. Should the model focus on previous job titles? On skills listed in the CV? On something more subjective, like the way someone writes? Every choice embeds assumptions, and those assumptions will quietly shape who gets shortlisted and who gets ignored. It is tempting to push these problems onto a future committee, or to assume that legal teams will catch them later. In practice, by the time legal looks at a system, the most important decisions have already been baked in. The good news is that asking better questions doesn't require a PhD. Three simple questions go a long way. Who might be harmed if this works as intended? Who might be harmed if it doesn't? And whose perspective is missing from this meeting? If everyone involved in building AI tools made a habit of asking those three questions out loud, we would probably avoid most of the headlines that make AI sound scary. Ethics is not a final inspection. It is a craft that has to be practised, line by line, decision by decision.