Universal Basic Salary: Help employers train workers. Drop failed programs.

A response to Will Raderman and Gabe Menchaca.

AI is our opportunity to rewire workforce training.

Will Raderman exhorts us to be flexible, and also asks us “to move on from programs that fail to produce results.” He insists that we follow the evidence and the data.

But if we do follow the evidence and the data, at this point what we should be “moving on” from are some of the other things Will suggests, such as programs giving advice on crafting a resume, career guidance programs, and training provided by third parties. A quasi-experimental study conducted by Mathematica and the Urban Institute found that apprenticeships gave workers an extra $12,920 to $18,772 per year compared to comparison groups of community college students and beneficiaries of career counseling and job search programs. A massive series of studies by the Department of Labor examined apprenticeships from many angles; one of the findings was that employers receive $1.44 for every $1 they invest in apprenticeships. Meanwhile, though programs to help people with their resumes can get them back into the labor force more quickly, they have no effect on long-term trajectories. Indeed, even the study Will cites on short-term credentials suggests that two-thirds of them have little value, and this is using an expansive definition of value.

If we follow the data, we come right back to apprenticeships. Where I agree with Will is that we should avoid trying to predict which industries should get more apprentices. We need an open, flexible, all-of-the above approach to apprenticeships. We need to make it easy for any firm that wants a registered apprentice to hire one, and we need programs to promote the option to firms and even to would-be firms. We need to figure out how to get very small companies to take on apprentices, and we need to think about apprenticeships for startups and apprenticeships for founders. Indeed, Gabe Menchaca’s delightful essay describes what are essentially apprenticeships for spies. But Gabe doesn’t tell us how to get companies in the private sector to focus on the long term and experiment with such models, especially if we see a raft of very small companies arriving. The system of registered apprenticeships is a proven way of doing so and a foundation for experimentation with new forms.

Gabe does put his finger on the main flaw of the apprenticeships-for-AI-dislocation model, which is that AI threatens to wipe out entire industries. What industry credential can protect against that?

There is a way to mitigate this problem, but apprentice pay would need to rise and apprenticeships would need to come with decent benefits. The idea is that when AI wipes out the legal career you trained for, and you are in your late 20s with hopes of starting a family or in your late 30s with a mortgage to pay off, an apprenticeship gives you a way to pay your bills while retraining to become a therapist. And if therapists are also wiped out by AI? Then you do the same thing over again with some other industry, perhaps one of the new industries being created by all the money that companies are saving by automating elsewhere.

If an apprenticeship position comes with decent pay and retirement savings and health benefits, there’s no reason you couldn’t do it as many times as necessary. If this can be done, then apprenticeships essentially unlink “training” from the particular phase in life when people can afford to have low earnings. Think of it as a universal basic salary: more mathematically plausible than universal basic income, because firms are sharing the costs and getting work from the apprentices in return. But unlike universal basic income, we could get there incrementally, starting from where we are—bipartisan agreement on apprenticeships—and building from there.

Note, by the way, that nothing in what I have said excludes the possibility of also raising taxes on profits, or developing more robust systems of welfare provision and unemployment insurance. Indeed, it’s quite the opposite: Everywhere we see a developed welfare state, we see a thriving system of apprenticeships. Really the question is not whether to move forward with apprenticeships for AI dislocation, but how.

---

Monica Prasad is Bloomberg Distinguished Professor of Economic and Political Sociology at the Johns Hopkins University and a Senior Fellow at the Niskanen Center.

---