Since my first review, AI models have traded real money, beaten market indexes, suffered serious losses—and given us a much stranger answer to the question of whether machines can trade.
On January 12, 2026, six leading AI models received $10,000 each and entered the prediction markets. They could search the web, read contract rules, compare prices, calculate position sizes, and place trades without human approval. This was no longer the familiar demonstration in which a model explains why it likes a particular stock.
A year ago, patients were turning ChatGPT into an unofficial medical assistant on their own. In 2026, that behavior became a product: AI can connect to medical records and wearable devices, influence the decision of whether to see a doctor, and in some services already route patients into the real healthcare system. The doctor has not disappeared. But the space in front of the doctor’s door is changing hands fast. In the summer of 2026, Florida pastor Scott Winters sued OpenAI.
AI agents are already doing work that, until recently, was done by people. Some companies are cutting staff, others are simply no longer opening new roles, and some are discovering that digital labor can be surprisingly expensive. Meanwhile, the people who keep their jobs are increasingly becoming operators and supervisors of machines. I tried to understand what is actually happening with the economics of AI agents by the end of summer 2026 — and where this story could go over the next nine months.
Until very recently, a good idea had one rather inconvenient characteristic: someone had to build it.
Suppose you’re a marketer. You understand the market, you know where to find your first customers, and one day you notice a problem people seem willing to pay to solve. Great. Now you need a developer. Then a designer. Then you discover you need another developer because the first one is working on the backend, while the app, inconveniently, also needs a frontend.
Generative AI improves essays, code, and grades faster than universities can work out what a piece of student work now proves — and what stays with its author once the laptop closes.
Picture two students handed the same assignment. Both are bright, both are motivated, both want the top grade — and from there their paths split.
The first loves the subject. She reads the sources, spends a long time building her argument, writes a messy draft, and only then opens ChatGPT — not to replace her thinking but to stress-test it.
Mass unemployment hasn’t arrived. But the claim that “nothing has changed” no longer holds up either.
Five years ago, most people answered the question “should I go into tech?” the same way: yes. Learn Python, build a few projects, land your first job, gain experience, grow. It wasn’t a guarantee, but it worked well enough that millions of people built their plans around it. Today the answer isn’t so obvious — not because developers are no longer needed, but because nobody can honestly say what that first career step will look like in three or four years.
AICaver
