Behind the scenes in Washington, many policymakers are planning for an economic disruption: potential job displacement caused by AI.

Yet some act like today’s safety net doesn’t exist, despite the massive $1.2 trillion a year in federal funds spent on 90-plus means-tested programs.

Instead of reforming programs, they’d create new ones — including one resembling a deeply flawed pandemic program riddled with fraud.

We have to do better than that.  

A New York Times article warning that “the federal safety net isn’t ready” for the potential shock of “widespread job losses” attributable to AI is a good example of such misguided thinking.

And it’s consistent with widespread fear-mongering about “AI’s potential to wipe out jobs, eradicate privacy and possibly even endanger humanity.” 

Despite such apocalyptic rhetoric, the extent of AI-driven job losses is unknown. Unemployment remains historically low, and poverty, when properly measured, has trended downward for decades.  

That should be cause for policymaking humility, not bold ambitions. But federal policymaking is anything but normal.

As Rahm Emanuel famously said, “You never want a serious crisis to go to waste” — especially when it’s an opportunity to realize your vision for a bigger, more expensive federal government.

For liberals, that means AI fears offer a chance to create more programs, instead of asking whether the current leviathan should be reformed to better meet needs.

Sen. Ron Wyden (D-Ore.), notably, proposes reviving a pandemic-era program he authored that provided a vastly expanded range of unemployment checks to non-workers and others not covered by unemployment insurance. 

Wyden’s no ordinary member; he’s the senior Democrat on the Senate Finance Committee, which oversees taxes and vast swaths of federal spending, including on unemployment checks.

If Democrats sweep this fall’s elections, he’ll once again play a key role in determining who collects future unemployment and other benefit checks.

Nor is what he proposes an ordinary program: It’s a redo of the deeply flawed Pandemic Unemployment Assistance program.

PUA expired in late 2021, but while it operated it was one of the most fraud-prone programs in history, with a staggering 36% improper-payment rate.

An Illinois audit found 50% of PUA checks were stolen by identity thieves. Colorado reported 75% of claims were fraudulent.

California found 95% of confirmed unemployment fraud there involved PUA. 

Overall, the program and other temporary federal unemployment programs lost more than $200 billion to improper payments and fraud, very little of which has been or ever will be recovered.

Since PUA expired, Wyden has called for reviving the program and its expanded benefit spending on a permanent basis.

Guided by the “Emanuel rule,” he’s using the threat of AI-induced job losses to make the case.

But that requires ignoring the experience of those who ran the troubled pandemic program.

As Robert Asaro Angelo, New Jersey’s labor commissioner, testified in 2023, PUA created a “perfect recipe for fraud,” and he advised lawmakers: “Don’t ever pass a program like PUA again.”

Instead of reviving PUA, today’s unwieldy safety net needs to be modernized and streamlined: Dozens of outdated programs that add unnecessary complexity need to be consolidated.

The oldest of 90-plus means-tested federal programs date from the New Deal of the 1930s, with major additions starting with the Great Society of the 1960s.

None, obviously, was designed with AI-induced job challenges in mind.  

That creates an opening for realistic reformers.

As my colleague Angela Rachidi and I sketched out in the new American Enterprise Institute volume “Land of Opportunity,” lawmakers could merge 70 small and medium-sized safety-net programs — with over $115 billion in annual funding — into a flexible block grant to states, which states could use for retraining, relocation or other important needs.

If AI does lead to significant job loss, states would be far better positioned to direct these resources to swiftly aid affected individuals, while ensuring that capped federal funds are deployed effectively.

In contrast, reviving an open-ended federal entitlement would only add to our already growing debt and undermine state accountability for program integrity and effectiveness.

Policymakers must do better than that.

Matt Weidinger is a senior fellow and Rowe Scholar in opportunity and mobility studies at the American Enterprise Institute.