fund of $55bn or so. If the stake earned a healthy 10% a year, the fund would grow to $140bn or so after a decade. Paying out 4% annually, a rule of thumb for preserving a fund in perpetuity, would amount to $20 a year per American annually. To be sure, the point of such schemes is that AI may prove anything but ordinary. Yet even if it does change everything, the firms’ combined value rises ten- or 20-fold and the state picks the right winners, Americans would get a yearly payout of a few hundred dollars. This is not enough to make anyone rich. Since it is anyway unclear where AI’s rents will ultimately accrue, this argues for looking across the industry. An annual levy of 0.2% of market value, along the lines proposed by some advocates of a broad wealth tax, could raise roughly $40bn a year at the current valuations of AI labs, chipmakers and cloud providers. But deciding what counts as an AI firm— and how much of Amazon, Google or SpaceX does—would be contentious. And the resulting dividend would still be at best a few hundred dollars per American annually. A nice bung—but well short of a universal basic income or meaningful insurance against widespread job displacement. Disbursing the proceeds involves more choices. One model is Norway’s oil fund, whose returns help fund public services. Another is Alaska’s Permanent Fund, which invests the state’s resource revenues on behalf of residents and pays them annual dividends (similar to what OpenAI and Anthropic have proposed and we assumed in our calculations above). A third is something akin to the “Trump accounts” the president has proposed for American tots. These would be seeded by the state, compound over time and be used to fund college or a pension, say. The left may prefer the Norwegian way. Mr Trump would favour one of the other two. Public ownership carries risks. It blurs the line between regulator and shareholder. Politicians may be unwilling to pursue antitrust action or impose costly safety rules on publicly owned firms, and happy to prop up stumbling ones (and their valuations). This may entrench incumbents and weaken competition. AI rents may also flow somewhere unexpected: electricity transformed society, but a wealth fund built around electric utilities would be a dud.
The best way to minimise such downsides may be a public wealth fund invested in a broad equity index. The capital could come from taxes on AI profits or mandatory equity contributions from across the AI economy— which, if the optimists are right, may one day mean most business. Even extraordinary returns would still probably translate into a modest dividend, arriving years from now. The harder questions—how to tax AI, how to regulate it and how to support displaced workers—would remain.■ Subscribers to The Economist can sign up to our Opinion newsletter, which brings together the best of our leaders, columns, guest essays and reader correspondence. This article was downloaded by zlibrary from https://www.economist.com//finance-and-economics/2026/06/11/how-to-share-ai-riches
How artificial intelligence got better at building itself New techniques can predict and prevent lung cancer Too much Chinese science is ignored by the West The chemicals that reduce wrinkles
Science & technology | Over and over How artificial intelligence got better at building itself What does “recursive self-improvement” mean for the technology? June 11th 2026 WHEN ANTHROPIC, an artificial-intelligence lab, debuts on stock markets later this year, it is likely to be one of the biggest initial public offerings in history. That’s because Claude, the company’s chatbot is beloved of coders, who are willing to pay a lot for access. Since Claude Code, its software- engineering agent, launched in February 2025, it has become indispensable for developers around the world. That includes Anthropic’s own: more than four-fifths of the code it published in May was written by Claude, the company says. Before Claude Code, the percentage was “low single-digits”. The systems have improved in quality of output as well as quantity. An influential benchmark from METR, a think-tank, shows that in early 2025 Anthropic’s models could complete tasks that took human engineers a little under an hour. The company’s latest systems can complete tasks that would take more than a working day. And so it may be easy to raise a cynical eyebrow when the company, at the top of its game and outclassing the competition, calls for the world to have “the option to slow or temporarily pause frontier AI development”, as it did on June 5th. What market leader would not wish that its competition stop trying to catch up? Yet Anthropic’s leaders, who have for years worried about the prospect of out-of-control AI wreaking havoc, seem sincere. The latest generation of AI models are such competent coders, engineers and (soon) scientists that many worry they may be among the last ever made by humans. Jack Clark, an Anthropic co-founder, thinks there is a 60% chance that, by the end of 2028,
an AI system will be capable of creating its own successor with no human involvement at all. That moment would mark the beginning of a process called “recursive self- improvement” (RSI), a closed loop. Version one of a model produces version two, which is faster and more capable; version two produces version three, which is more so again. The loop continues, and the improvements grow with each iteration. Build an AI system capable of this, and your human engineers never need to build another one again. “What can seem to many like a fanciful story may instead be a real trend,” says Mr Clark. Nobody knows for sure what the consequences of recursive self- improvement would be. Because AI can, unlike humans, work tirelessly and constantly, some think it would in short order lead to a superintelligent AI— a “fast take-off”. (It has also been onomatopoeically dubbed “going foom”, for the sound one might imagine an intelligence explosion making). AI doomers fear the superintelligence would be beyond human control, and that the start of RSI is the moment at which humanity’s fate is handed over to the machines. Yet a self-improving AI would probably face speed limits, at least at first. Building a model capable of RSI would require automating a range of specialist tasks currently carried out by humans. At present data scientists work on the theory of AI and coders put it into practice. Systems engineers build the foundations on which toy models can be raised to production scale. Other people seek out novel sources of training data, or experiment with ways to generate it fresh. Alignment and safety teams check that what comes out of the training process won’t cause harm, intentional or otherwise. Not all of those teams are equally amenable to AI assistance, and within each specialism some tasks are more automatable than others. It will not be too long until a human coder can do their job without ever writing a line of computer code themselves, but it may be some time until an AI is able to negotiate to acquire a previously undigitised collection of scientific papers. It is not always obvious how the “jagged frontier” will progress. Designing new algorithms seemed one of the safer jobs, until one of Google DeepMind’s models, AlphaEvolve, began doing it in May 2025. It proposed
a change to how Google spreads workloads across its data centres that saved 0.7% of the company’s worldwide computing power, and found better ways to perform matrix multiplication, which speeded up the training of Gemini, the company’s flagship large language model (LLM), by 1%. Full RSI requires every task in this chain to become automated. The AI- powered acceleration of research and development (R&D) may be felt before then, however. “As the fraction of AI R&D performed by AI systems increases, the productivity boost over human-only R&D” could increase ten- fold, then a hundred-fold, then a thousand-fold, according to a report published in January by the Centre for Security and Emerging Technology (CSET), a think-tank within Georgetown University. In that scenario, it warns that even if some aspects of AI R&D are initially difficult to automate, “the accelerated rate of progress means those bottlenecks are soon overcome.” Today no AI model can build its own successor. But big AI models can build smaller models on their own. With human help they can build other big AI models, too. Earlier this year Andrej Karpathy, a then-independent researcher who now works for Anthropic, trained a chatbot about as capable as GPT-2, a large language model built by OpenAI in 2019. Back then the model took 168 hours of training to build on 32 state-of-the-art chips; Dr Karpathy achieved the same result using a single computer with eight GPUs, the specialised chips used to build AI, in only three hours. With some more months of work he reduced the training time for his model, Nanochat, to just over two hours. In March he handed the work of speeding up the training process over to an AI agent called Autoresearch. In two days the training time dropped to one hour and 48 minutes, and five days after that it fell to one hour and 39 minutes. “I didn’t touch anything,” Dr Karpathy says. The 18% improvement on the human work is striking because Dr Karpathy is a particularly talented human: he was a founding member of the research team at OpenAI and the head of AI at Tesla for five years. The improvements themselves were prosaic. The AI agent picked better starting values for the training run, widened the scope of the LLM’s “attention” window and noticed that the model’s focus was wandering. None