There's too much leading from behind on AI and workforce.
Here's what we should be doing instead.
The issue.
We can do better than wait for somebody else to come up with a plan for handling potential job loss from AI.1
Explain.
Something big is missing in the media reports I have read about the threat AI job loss. In between quotes from tech CEOs—or in bolder circumstances, carmaker and even retail CEOs—there really isn’t a whole lot of specificity on anything should do to prepare for the jobs of the AI age. Just that their livelihoods are about to get hit by an automated truck.
From Axios over the weekend:
“Artificial intelligence is going to replace literally half of all white-collar workers in the U.S.,” [Ford CEO Jim Farley] told author Walter Isaacson at the Aspen Ideas Festival. “AI will leave a lot of white-collar people behind.” . . .
Amazon CEO Andy Jassy said in June that Amazon will reduce headcount “as we get efficiency gains from using AI”: “It’s hard to know exactly where this nets out over time, but in the next few years, we expect that this will reduce our total corporate workforce as we get efficiency gains from using AI extensively across the company.”
There’s also, um, this from August.
[OpenAI CEO Sam Altman] said that the coming decade would be the most exciting time in history to start a career, especially those who dream about working in space. “In 2035, that graduating college student, if they still go to college at all, could very well be leaving on a mission to explore the solar system on a spaceship in some completely new, exciting, super well-paid, super interesting job,” he said. He added that they’ll also be “feeling so bad for you and I that we had to do this really boring, old work and everything is just better.”
Neat.
The Axios piece also gets to a complaint I have heard a fair amount over the past few months:
The White House and Congress aren’t treating this like a brewing crisis, and seem more focused on beating China to advanced AI than bracing workers for a short-term jolt.
That point is not wrong, but speaking from experience inside government, I don’t think the inaction is because the problem is being ignored. In fact, everything I have seen and heard is that the Administration is treating AI job loss as a very serious issue.
Rather, I think the situation is a little more nuanced and a lot more frustrating. Because this is yet again where the White House’s rigid politics prevent it from doing what the situation needs—particularly since a reason all those CEO comments seem kind of light on substance is they likely are nowhere close to figuring out these big questions.
Defer. Defer. Defer.
I’m a big fan of reaching out to employers and partnering with them to understand what they get out of a workforce program. Evidence continues to suggest employers are the ones who do all the hiring. Ergo, we should involve The Hiring People when we’re training workers to get hired. But deference, alone, has its limits.
And yet, deference seems to be almost the whole of the Trump strategy on AI and other issues. For example:
Something I have quoted many times now is Deputy Labor Secretary Keith Sonderling telling corporate executives in June that they “need to tell us where those jobs are going” on AI so they can “so you have those skills you need and the workers are ready to go.”
In July, the White House’s AI action plan made broad recommendations for federal agencies to “convene employers, industry groups, and other workforce stakeholders to develop or identify national skill frameworks and competency models for these roles.”
In August, the Trump workforce blueprint insisted that “[t]he input and direction of these companies on how the workforce system evolves to meet their needs is critical for the success of the Trump Administration’s workforce strategy.”
At the end of August, the Trump II DOL published guidance on how to use federal funding to address and prepare for AI’s impacts on the workforce. Most of the feedback I have heard—and can’t wholly disagree with—is that it’s kind of empty. The guidance provides very little… guidance on what are the best ways to spend money to prepare workers for AI jobs and AI job loss. Rather, it mostly identifies pots of money that didn’t need that much decoding to be wired toward AI.
To the extent the guidance said something, it has emphasized “AI literacy.” I wrote in July about how that term, as used by tech CEOs, isn’t really a job skill so much as something someone can pick up in an afternoon, thanks to the user-friendliness of AI interfaces. It’s not a concept to wholly dismiss, but it’s definitely become something of a gap-filler for business leaders who want to seem like they’re saying something about AI when they don’t have anything terribly concrete to say in fact.
Which speaks to the bigger problem: if employers are still figuring out what AI is going to do to the workplace, what exactly is there to defer to?
What is an AI job—or AI job loss—anyway?
As I wrote in July, if you’re looking for jobs created for AI, they’re probably pre-existing roles remade around integrating AI as a tool for doing the work better—since that is what AI is at this stage. There also is a time lag here clouded by the hype and speed of AI development, but seems fairly obvious if you think about it. Emerging occupations need to emerge, and it’s very hard for employers to say what skills they need to fill them if they are still emerging them.
In that case, it makes sense to be aware, but patient, and not hurl a bunch of resources too soon in one direction. I really, really don’t think the Trump Administration is wired that way. If it wants an Obama-era example for why it should be patient, think of the failed effort to turn miners into coders, an idea that didn’t really pan out. If it had, it likely would have routed people into another round of job losses, since coding is one of the jobs with the most tangible job losses due to AI.
Which gets to another thorny issue: creating or keeping jobs via AI is not what many employers seem to be thinking about. Much of the pitch of AI is that employers don’t need as much staff as they need now because of AI. Heck, it’s something I feel like I have gotten in my inbox twice a week since I put “CEO” in my title on LinkedIn, and the only person I employ is me.
Even if companies are not firing a bunch people because of AI, they seem to be saying they are. From the Lightcast report I mention above:
There’s growing evidence that many high-profile layoffs announced with references to AI are in fact examples of routine economic cooling or efficiency gains from technology—processes that have taken place repeatedly over past economic cycles. In many cases, referencing AI is a convenient explanation for cuts that would have happened regardless of the technology’s impact, or even a strategic tool for demonstrating nimbleness to investors. Closer analysis of layoff data, company statements, and hiring patterns suggests that while some jobs are certainly being reshaped or eliminated by AI, many more reductions fit long-standing business trends, with AI serving as the latest rationale presented to the public.
On that note, recent research by MIT indicates that the overwhelming failure of employer efforts to replace workers with AI. That report suggested most of the gains in productivity coming from increased work by shorthanded staffs.
So what do we do about it?
Let’s talk about what I think the Administration will do in the near term and what somebody should do.
My expectation is that the Trump Administration will put out dedicated AI workforce money—and probably soon. I don’t think this will be for “AI jobs" but what this summer’s White House action plan called “AI infrastructure jobs”—trades jobs like electricians and HVAC specialists needed to successfully build and operate data centers. I don’t think that will be satisfying to the people bothered the White House isn’t serious about this issue, but in terms of doing something in light of the White House’s political constraints, I think it’s a good idea.
So what should it be doing? Well, something that I think is hard to build for any administration, but this one is really not built for because it could be taken as government telling employers what they should do: a regularly updated resource that gives direction and substance, not platitudes, on
What occupations and companies need to be aware of near-term AI job loss.
What workers can do to prepare for that AI job loss and what federal resources are available.
What “AI jobs” are available, what skills they need, and how long these jobs—and the skills—are likely to stick around.
In short: something that helps out employers by removing some of the pressure of needing to explain what they need or translate it into terms outside their enterprise. It’s also the type of thing that can be built in a way that it can be a freestanding resource for groups applying for federal funding—and private funders in deciding and building their opportunities. The Administration just tried this with a Commerce grant and its workforce blueprint, but it’s too vague—and too focused on why federal workforce funding isn’t built the way it likes—to be useful for that purpose.
The challenge is that the hype and stakes of AI really makes it hard to get what you need to build a document that is useful. This can’t be a Big Task Force with Big Names Who Make Commitments, or a Big National Convening like what’s in the White House AI action plan, but something based upon—and translated from—the actual and relatively unvarnished thinking of employers. That’s better handled by a small group of policy professionals who don’t have the weight of a title that can limit what feedback they get and drive employers to have Someone Important to do the talking even though Someone Important doesn’t have the details to build this kind of project.
Again, this is really isn’t the Trump Administration’s thing, and the Minimal Acceptable Levels of Bluster necessitated to satisfy the White House definitely interferes with the active listening need to figure this stuff out. The sad part is that I think the federal government is the best entity to do this type of work. It has the most reach and resources, the east amount of conflicts of interest, and the greatest capability to marshal attention that actually gets someone to use the final product.
Somebody should do it, though. With all due respect, I think waiting around and hoping for that dope AI job on Jupiter is not the most effective strategy for American workers at the moment.
Card subject to change.
I needed to pivot the topic today, but next week, I’ll get to the piece I talked about last Tuesday about what workforce programs really need and all workforce politics seems capable of producing.
I think we’re heading for a shutdown and a certain degree of carnage associated with it. In all likelihood, I’ll be back midweek with a shorter piece on it for paid subscribers, but if things evolve too quick, I’ll hold until Friday.
Either way, Friday I’ll clean up what federal money just didn’t go out and what money I would be worried about if I had competed for it. Looking at you, Women in Apprenticeship and Nontraditional Occupations.
See y’all later this week.
An editorial note: one of the things I have heard from readers in recent months is a sincere wish that JTW not do much—or any—AI coverage due to some general exhaustion with it. I definitely can’t say that I won’t write anything AI—see pretty much everything in the piece above. But my siblings, I definitely get it.
When I write about AI in this space, I’m leaning as far away as I can from the binary of “AI Is Going to Make Everything Better Including Hats Do You Wanna See My AI Hat?” and “AI Is Coming to Take My Job and My Sweater Claude Is Wearing My Sweater Right Now” and toward specificity and tangibility for how we’re reacting to it and explain how we can make better workforce policy in the process.