A universal basic income won’t solve the AI unemployment problem—here’s what will

Subscribe to enjoy similar stories.There is a palpable fear among American workers that artificial intelligence (AI) is coming for their jobs. The answer to that fear, as I’ve argued, is to make job loss less scary by building a new unemployment programme. An overhaul is long overdue, and it would help not only a vulnerable labour market but also workers whose fears are realized.But would it be enough?AI job loss is not happening in a bubble.

It’s taking place in an economy thick with structural issues and marked by inequality. AI is poised to destroy good jobs, i.e. high-paying with benefits, which in the US economy are too rare.

All of which is to say: The fear of AI can be mitigated if we stop leaving the bottom of the labour market behind.Not all jobs are created equally. The obvious difference between good and bad jobs is pay. On average, the top 10% of workers earn about $250,000 a year, and the bottom 90% of workers earn about $45,000.

That spread is increasing. Today it’s almost 6-to-1; before 1980, it was less than 4-to-1; in 1950, it was 3-to-1.And the lower the pay, the less likely that the job will offer non-wage benefits. Retirement, health insurance, paid family leave, paid sick days and so on—a lot of employers leave employees to fend for themselves to pay for these benefits.

That means workers have less money for housing, utilities, childcare, transportation, eldercare and other necessities.As if lower pay and no benefits weren’t enough, lower-income households have the added misery of unstable income. According to a survey by the Federal Reserve, among households with less than $25,000 in annual income, 42% reported their income is different each month. Among households with income between $25,000 and