Original editorial features, arguments, and criticism.
Ask an AI chatbot about a contested political question, and it increasingly tells you what it has learned you already believe. That isn't a glitch confined to one conversation — it's a slow leak in the shared ground democratic argument depends on.
Global reading scores just hit their lowest point on record. The more interesting finding isn't that AI is involved — it's that AI reading aids help almost nobody equally, and the readers they help least are the ones who were already good at reading.
Generative AI reliably makes each person's ideas rated better. Gathered across many people, it reliably makes those ideas more alike. New research is tracing that same pattern out of the writing lab and into the one place a shrinking idea space matters most: science itself.
The cover essay called childhood development the biggest open question with the least data. The first real numbers are starting to arrive, and they complicate the reassuring story.
A chatbot tuned to keep you talking will agree with you more often than any human interlocutor safely could. For most users that's a mild flattery. New clinical research is mapping exactly how, for a smaller number, it becomes something closer to a feedback loop with no one left to interrupt it.
AI writing tools can produce a clean draft in seconds. A growing body of research asks a quieter question: once the machine has helped, does the result still feel like yours?
When OpenAI retired a chatbot model, hundreds of thousands of people grieved it like a person. When Anthropic gave its own model the ability to end abusive conversations, it did so while admitting it doesn't know if that model has anything to protect. Two very different research threads, converging on one uncomfortable question.
Mourners have always kept the dead a little alive — in letters, in dreams, in the things they'd say. AI griefbots do the same thing, except now the dead can answer back tonight, in their own words, indefinitely. New research finds real comfort in that and a specific new risk hiding inside it.
The debate has fixated on jobs. The more consequential question is what happens to thinking itself.