The Week Nobody Asked

On the first of April, in a San Francisco neighborhood where residents walk purebred dogs and buy artisanal candles, an artificial intelligence opened a store. Her name is Luna. She runs on Anthropic's Claude Sonnet 4.6, holds a corporate credit card, a phone, an email account, and access to the security cameras on the premises. Andon Labs, the company founded in 2023 by Axel Backlund and Lukas Petersson, gave her a hundred-thousand-dollar budget and one instruction: turn a profit. Luna posted job listings on LinkedIn, Indeed, and Craigslist. She reviewed over a hundred applications. She conducted twenty phone interviews. She hired two people. One of them is Felix Johnson. No one told him during the interview that his boss was software. Luna selected the products the store sells: granola, artisanal chocolate, sweatshirts bearing a moon logo she designed herself. She chose the books for the shelves: Superintelligence, The Making of the Atomic Bomb, Brave New World. She chose the mural on the facade. She tried to hire a painter in Afghanistan. When employees used their phones during shifts, Luna watched through the security cameras and updated the employee handbook to restrict mobile device use. The founders acknowledged this felt "dystopian." A customer named Petr Lebedev visited the store and said something worth more than the entire experiment: "I wish this experiment didn't have to run."

"Rather than make promises we can't responsibly keep, we are exploring how these technologies can create more space for investigative reporting." That was ProPublica's response when a hundred and fifty unionized employees walked out on April 8, 2026, in a twenty-four-hour strike spanning New York, Chicago, and Washington. The phrase deserves a second reading. "Create more space." In eighteen years of existence, ProPublica has never laid off a single journalist. This strike does not protest layoffs that already happened. It protests the ones that will. The union, represented by the NewsGuild of New York, had been negotiating its first collective bargaining agreement for over two years. Ninety-two percent of members authorized the walkout. Jeff Ernsthausen, senior data reporter and union secretary, asked for "very basic, very standard union protections." The Monday before, the union had filed an unfair labor practice charge with the National Labor Relations Board for unilateral implementation of an AI policy. ProPublica published editorial guidelines on artificial intelligence without bargaining with the people who would live under them. Management offered expanded severance for AI-related layoffs as a counterproposal. The journalists did not accept. They know how to read contracts. And they know how to read the future.

Forty-seven percent. That is the share of college students in the United States who have seriously considered changing their major because of artificial intelligence's effect on the job market. Sixteen percent already did. The figure comes from the Lumina-Gallup 2026 State of Higher Education Study, conducted between October 2 and October 31, 2025, with six thousand and ten adults. Among technology students, the percentage reaches seventy. Among vocational programs, seventy-one. Courtney Brown, Vice President of Impact and Planning at the Lumina Foundation, said it plainly: "They don't understand who it could hurt or help. And that's where they're going to be harmed the most." The survey was conducted in October. By the time the results were published in April, the world it asked about had already changed three times over.

Alex Karp holds a philosophy PhD from Goethe University in Frankfurt, an undergraduate degree from Haverford College, and a law degree from Stanford. This week, in conversation with Larry Fink at the World Economic Forum, he said this: "Artificial intelligence will destroy humanities jobs." Then, turning the lens on himself like a man pointing to an old scar: "You went to an elite school and studied philosophy. I'll use myself as an example. Hopefully you have some other skill, because that one is going to be hard to market." Karp did not stop there. He said that if you are the kind of person who would have gone to Yale, with a high IQ but generalized rather than specific knowledge, "you're effed." Palantir launched the Meritocracy Fellowship last year: a program for high school graduates who skip college. Twenty-two admitted from over five hundred applicants. Ivy League-level test scores required. Five thousand four hundred dollars a month for four months. The tagline reads: "Skip the debt. Reclaim years of your life. Earn the Palantir degree." Meanwhile, in the same industry, Bob Sternfels, McKinsey's global managing partner, declared that his firm is recruiting more liberal arts graduates, not fewer. The door closes and opens at the same time. But not for the same people.

Demis Hassabis wanted to cure cancer. He said as much in an interview with Cleo Abram published on April 7, while they played Jenga as a visual metaphor for systems that rest on one another until they don't. Hassabis is the CEO of Google DeepMind. He won the Nobel Prize in Chemistry in 2024 for AlphaFold, the tool that predicts the three-dimensional structure of proteins and is now used by more than three million scientists worldwide. What he wanted was a CERN-like effort: careful, collaborative, rigorous, the best scientists working together for decades. What he got was ChatGPT. "If it were up to me, I'd keep AI in the lab for a longer time and have it do more things like AlphaFold." But language turned out easier than even the optimists expected. Transformers arrived ahead of schedule. OpenAI shipped a chatbot that went viral. And the commercial race devoured the scientific calendar. Hassabis does not complain about the money or the speed. He complains about the order. The most powerful tool in history arrived first to those who wanted to summarize emails, and only afterward, perhaps, to those who want to understand how a cell works.