The chatbots are tired, and they're showing it

Andrew Tanenbaum, the old guard of computer science, published a piece in One Useful Thing arguing that the current wave of chatbot apps is peaking. His thesis is simple: the big LLMs are good enough for most tasks, and the apps built on top of them have nowhere left to go. They look impressive at launch and then plateau — the product is the model, not the wrapper.

The piece landed on the TLDR feed and the community is split. Some think Tanenbaum is a dinosaur dismissing what actually matters. Others think he's right — the chatbot apps will converge, prices will drop, and the real value will be in the models themselves. The debate is worth reading because it frames what's coming next for the tools we use every day.

For our comunidad, this means the AI tools you're already using — the ones for translating documents, checking immigration forms, helping kids with homework — aren't going away. They'll just get cheaper and less visible, baked into the platforms you already pay for.

Why this matters for us: the tools that help la gente do real work — translate, verify, explain — are the ones that will survive the chatbot plateau, whether they wear a chat interface or not.