AI and People
What AI can and cannot do, from assistants to general intelligence.
Start here · Essential guide
What artificial general intelligence would mean compared with today's narrow AI, the main debates about whether it is reachable, and why timelines are so uncertain.
A thought exercise surveying five broad areas of AI research, from reasoning to robotics, with no labs or products named and no experiments of our own reported.
Which routine tasks AI assistants can plausibly speed up, where they fall short, and the hidden costs such as checking output that can offset time savings.
How customer service chatbots work, where they help, where they frustrate people, and when a conversation should be handed to a human agent.
The social and emotional limits of voice and text assistants compared with people, and why convenience is not the same as connection or care.
A plain-language explanation of machine learning: how programs learn patterns from data instead of following hand-written rules, and where that approach goes wrong.
How layered networks of simple units process information, how they are trained by adjusting connection strengths, and why they work well yet remain hard to interpret.
A thought exercise on trolley-style dilemmas and fairness questions for machines that make decisions, noting that no experiment was run and no answer is settled.
How spoken audio is turned into text by speech recognition systems, from sound waves to words, and why accents, noise and homophones cause errors.
Related topics