As AI adoption accelerates, society is rightly asking how much we can trust AI systems. But maybe the first question should be, how did we ever trust human ones?We know that large language models, the technology behind tools like ChatGPT, can make things up. They find patterns where none exist. They lose track of information beyond a certain point, the way we struggle to hold too many things in our heads at once. They can reconstruct facts to fit a narrative when answering a question, smoothing over gaps in reasoning rather than flagging them.Subscribe to gift this articleGift 5 articles to anyone you choose each month when you subscribe.Subscribe nowAlready a subscriber? Fetching latest articles
AI is like aviation. It needs systems to guard against human fallibility
The issue is not whether artificial intelligence can be trusted but how systems are designed to manage its imperfections.











