Friends,
Anthropic — the artificial intelligence company behind the chatbot Claude —confidentially filed yesterday for an initial public offering. Soon, SpaceX and OpenAI will be doing the same. The three I.P.O.s could mark a once-in-a-generation creation of giant AI firms, and huge money for a relative handful of people (including the world’s first trillionaire, Elon Musk).
But this giant AI wave is also likely to spell the end of a huge number of jobs. AI will also create some new jobs but the AI tsunami is coming so quickly that, unlike in previous technological waves, most workers won’t have much chance to adapt.
Not only will AI destroy jobs, but workers will have even less bargaining leverage than they now have, which means that to stay employed (or land a job) they’ll have to settle for lower wages.
Put this together with (as I pointed out Sunday) languishing returns to labor and soaring returns to capital even before the AI tsunami hits, and AI is sure to shift the balance of economic power even further from labor to capital — making wages an even less reliable mechanism for distributing prosperity. We’re heading for a major shakeup of our entire political economic system — and it’s coming soon.
So what should be done?
Herewith, four ideas that are being actively discussed — in schools of public policy, in legislative cloakrooms, and among the billionaire class. I’m going to set each of them out as clearly as I can, along with their downsides, and ask you to weigh in.
1. Rely on charitable giving, especially by newly minted AI zillionaires.
Some see a “third wave” of American philanthropy emerging from AI. Anthropic and OpenAI have both pledged a large part of their shares to charity. More billionaires are likely to be created in the next three months than ever before. If they give away a portion of their gains, we may be witnessing a tidal wave of hundreds of billions in charitable giving generated by AI wealth.
But how can we be certain that the AI billionaires will contribute their wealth to the things America really needs? Too often in the past, charitable giving has come in the form of publicly subsidized (through the “charitable” tax deduction) investments in ego monuments to the wealthy — new wings of art museums, new Ivy League academic “centers,” and symphony halls, on which their names are engraved. And even if the giving is better aimed, how do we guarantee that they’ll give much of anything, anyway?
2. Create a universal basic income, financed by taxes on AI corporate profits.
A second idea is a universal basic income, financed by taxes on the profits of AI corporations or on billionaire wealth, or both, that would go to everyone and provide enough to meet subsistence needs. It would be roughly analogous to Alaska’s Permanent Fund, which distributes a portion of the state’s oil and mineral revenues in yearly payments to all Alaska residents.
But some argue that even a barebones universal basic income would discourage work; a subsistence wage will cause some to work fewer hours or none at all. Others argue, to the contrary, that meeting subsistence needs isn’t nearly enough; many families are already stretched to the limit, and when AI takes away their jobs or incomes they’ll need far more. Still others worry that a universal basic income might lead to further reductions in public assistance, such as Medicaid and food stamps, on the assumption that lower-income people’s needs are already being met by the basic income.
3. Create universal basic capital.
A third idea is that we not rely on government taxes and transfers, but create instead a system of universal basic capital. This would give every citizen a share in a national fund that includes partial ownership in AI corporations — shares whose value will grow as the nation’s (and AI’s) wealth grows. Shareholder-citizens would earn regular dividends and build permanent personal wealth over time. Such shares could take many forms: baby bonds that build capital from birth, direct equity stakes in publicly funded AI infrastructure, even allocations of compute capacity that holders can use or sell.
But critics worry that to create a sovereign wealth fund sufficiently large to generate meaningful returns to all citizens would require very large upfront public expenditures, redirection of existing public spending, and mandatory contributions of corporate equity — all politically difficult to pull off. Such a fund would also raise a host of difficult governance questions about how ownership rights are to be exercised, how assets are selected and diversified, and how to prevent political interference in investment decisions.
4. Enlarge Social Security so workers can retire at 50.
A fourth idea being discussed relies on the framework of the U.S. government’s most popular program — Social Security — and adapts it to an economy that won’t need nearly as many workers as before AI. It would allow workers to retire much earlier — say, at age 50 — with full benefits. Financing this would require a tax on AI whose proceeds went directly into the Social Security retirement fund, in a sufficient amount per year to keep the fund solvent at the lower retirement age.
But some argue that we shouldn’t tinker with Social Security. It’s already running out of funds — the latest projection shows it has just six more years in which it can pay out full benefits — and more demands on the program could destroy it. Others say that allowing workers to retire earlier would be regressive because higher-income workers live longer.
So today’s Office Hours question: Which of these ideas for dealing with the loss of good jobs to AI strikes you as most promising, despite their drawbacks? Feel free to add others.POLLHow to deal with the inevitable loss of good jobs to AI?Rely on charitable contributionsUniversal basic incomeUniversal basic capitalEnlarge Social SecurityOther (in comments)