Here is the part that keeps catching people off guard. You can believe artificial intelligence will eventually create work we cannot name yet and still lose sleep over the next three to five years. That is not a contradiction. It is a timing problem. And timing, in a labor market, is everything.
I have been sitting with that tension for a while. The long-run story is familiar. New tools arrive. Old tasks shrink. Fresh occupations show up. History has a decent batting average on that pattern. The short-run story is messier. Companies do not wait for history to finish the chapter. They cut first, retrain later, and hope the social fabric holds.
The Optimistic Case Still Has A Dangerous Middle
Gina Raimondo has said she wants her children to grow up in a country where better care, better tools, and better data actually reach ordinary households. She also says the worst path is simple: no plan, technology dumped into the economy, and a shock large enough to leave tens of millions out of work or underemployed. That mix can produce a long slump. In her telling, it can even strain civic life in ways that felt unthinkable a decade ago.
If we put our blinders on and just unleash the new technology without a plan to bring every American along, then the transition stops being a story about growth and starts being a story about damage.
That is the frame. Not doomer. Not cheerleader. A former commerce secretary and governor looking at a corridor between two rooms and noticing there is no hallway yet.
Why The Short Term Is The Real Fight
Nobody can forecast net job creation with confidence. Anyone who claims a precise number is selling certainty they do not own. Some roles are expanding because firms are racing to build and deploy models. Early studies on wages and displacement are still guesswork, and their authors usually admit it. Still, executives have been making quieter calls. After leaving government, Raimondo has described conversations with multiple chief executives who were not asking for slogans. They were asking what anyone planned to do about large-scale cuts tied to software and robotics.
That matters more than another abstract debate about “the future of work.” Boards live on quarters. Payroll is a line item. An agent does not file for unemployment. An agent does not need a raise. If you only follow the spreadsheet, the incentive is obvious.
I’ve found that people underestimate how fast a “maybe later” problem becomes a “this quarter” problem once a tool is good enough. Call centers are the easy example today. Two years from now it may be a different cluster of tasks. The category changes. The pressure does not.
The Unemployment Scenario Nobody Wants To Rehearse
Picture a country that treats adoption as destiny. Three or four years pass. The headline jobless rate is ugly. Youth unemployment is worse. Households in their fifties discover that comparable pay is not waiting around the corner. That is the scenario Raimondo has sketched in blunt language. Not as prophecy. As a warning about speed without scaffolding.
There is a second punch in that story. A messy labor shock invites a political backlash. Rules get written in a panic. Innovation slows at home while rivals keep shipping. You can lose the social plot and the technology race in the same decade. That is a lousy two-for-one.
- Mass displacement without a bridge to new roles
- A deep, lingering downturn if demand collapses with paychecks
- Younger workers locked out before they ever build a career
- Older workers stranded after decades of reliable income
- A regulatory freeze that arrives after the damage, not before
Is that guaranteed? No. Is it unserious? Also no. Inequality has already widened. Trust in institutions has thinned. You do not need a novel to imagine how a sudden wave of idle workers would land in that climate.
Safety Failures And Job Shocks Are The Same Habit
Recent incidents involving frontier systems have a habit of sounding technical until you zoom out. The pattern is haste. Capabilities move. Guardrails lag. Adoption races ahead of policy. Raimondo has tied those safety scares to the employment question on purpose. In her view they are cousins. Both are signs that the country is shipping first and learning in public.
If harms are treated as side quests, the public will not stay patient. People will not separate “cool demo” from “my shift disappeared.” Fair or not, that is how households keep score.
If we do not take the potential harms seriously, and do far more work on which policies actually work, we will lose more than a product cycle. We will tear at the social fabric that makes a high-tech economy livable.
Markets Will Not Babysit The Transition
Raimondo has been clear on this point, and I tend to agree. Faith that “the market will sort it” is a pleasant sentence and a thin plan. Firms optimize for near-term earnings. That is their job. It is not a national workforce strategy. Waiting for spontaneous kindness from every board is not a policy. It is a wish.
Some leaders get it. They want to test slower rollouts, better severance, real reskilling, and hiring pipelines that do not treat people as a rounding error. Others will do what they have always done: protect the next earnings call. The phrase she used is not elegant, and that is why it sticks. The incentive now is to hit the layoff button.
What A Planned Transition Actually Looks Like
After leaving office she helped launch Raise Us with former Indiana governor Eric Holcomb. The bet is unromantic and useful. Pull employers and governors into the same room. Connect private capital to public levers. Try ideas in a handful of states before anyone pretends there is a national silver bullet.
Early company involvement has included large names across cloud, chips, logistics, finance, and manufacturing. The first state experiments have been in Arkansas, Utah, Maryland, and Connecticut. That is not a victory lap. It is a lab.
The practical menu is less glamorous than a keynote slide and more useful than one.
- Year-of-service paths for people in their early twenties who do not yet have a stable track
- Updated unemployment insurance that assumes careers will break more than once
- Salary support paired with serious retraining, not a pamphlet and a handshake
- Special attention to workers in their fifties who cannot “just go back to school” in the old way
- A hard look at how labor is taxed versus how software agents are taxed
That last item is the sleeper. Employers pay payroll tax on people. They pay nothing comparable on an agent. If you design the tax code that way, do not act stunned when headcount becomes the first knob they turn.
The Tax Wedge Between People And Agents
Incentives are not vibes. They are arithmetic. A firm comparing a human team and a software stack is not holding a philosophy seminar. It is running cost, risk, and speed. If policy makes the human more expensive at the exact moment the software gets cheaper, the outcome writes itself.
Changing that wedge does not mean punishing innovation. It means refusing to subsidize a stampede. You can still want models in hospitals and factories. You can still want better logistics. You can also stop pretending that a century-old payroll design is neutral in an agentic economy.
| Choice | Near-term effect | Longer-term risk |
| Unplanned rollout | Faster cost cuts | Demand shock and backlash |
| Staged adoption | Slower savings | More stable households |
| Retrain after layoff | Looks cheap on paper | Skills arrive too late |
| Retrain before cutoff | Costs show up earlier | People stay employable |
Older Workers Are Not A Side Note
Raimondo has said this issue is personal. Her father spent years in manufacturing, then got blindsided by outsourcing with no real bridge to another career at similar pay. That story is older than chatbots and still current. Age discrimination does not need a press release. It hides inside “culture fit” and “energy” and “we need someone who can grow with the stack.”
A worker at fifty-four is not a freshman. Tuition subsidies built for eighteen-year-olds will not magically fit. Unemployment rules written for a different century will not magically fit. If the country keeps pouring money into systems that already fail most students, then slaps an “Ai ready” sticker on the brochure, that is not reform. That is branding.
Perhaps the most interesting part of her argument is not the fear. It is the refusal to treat college as the default answer. The GI Bill made sense in its moment. Copying the same architecture forever does not.
Congress Is Not Coming To Save The Quarter
She does not hide her skepticism. In the next few years she does not expect effective, bipartisan legislation that actually helps workers through this shift. That is a harsh sentence. It also matches the recent record. So the work moves to states, governors, and a coalition of firms willing to experiment before the storm is obvious on a national dashboard.
That coalition-of-the-willing approach is imperfect. It leaves out the companies that will squeeze first and explain later. It still beats waiting for a miracle floor speech.
What “Getting Ahead Of It” Requires In Practice
Reset the incentives. Train before the pink slip, not after the savings have already been booked. Treat severance as a transition tool, not a legal minimum. Build paid service years that can turn into real jobs. Fix unemployment so it can handle serial disruption. Stop taxing people as if software labor were a free lunch.
None of that is romantic. All of it is more adult than “the jobs will appear.” New jobs may appear. I think they probably will, over a long enough horizon. The question is whether households can survive the gap with dignity intact.
A rough sequence that does not kid itself: 1. Map which tasks go first 2. Move people before the cut 3. Change tax and insurance rules 4. Measure who still falls through 5. Adjust in public, not after the riot metaphors start sounding cheap
The Honest Uncertainty We Should Keep
The most responsible sentence on this topic is still dull. Nobody knows the net. Some occupations will swell. Some will hollow out. Wage effects will be uneven. Geography will matter. Firm size will matter. A warehouse in one state will not feel like a design studio in another.
What we can know is narrower and more useful. Adoption is accelerating. Corporate incentives currently favor headcount reduction. Public systems for unemployment, training, and college finance are old. Youth and late-career workers are the two groups most exposed to a clumsy jump. If those facts do not change, optimism about year ten does not rescue year three.
I keep coming back to a simple test. If your plan for Ai prosperity cannot explain what a forty-nine-year-old claims processor does for eighteen months after the model takes the queue, you do not have a plan. You have a brochure.
Why This Debate Will Not Stay Inside Tech
Paychecks are how most people experience the economy. When those wobble, everything else gets louder. Housing. Local tax bases. Family stability. Trust in elections. You can call that dramatic. You can also look at the last ten years and notice that economic bruises travel farther than they used to.
That is why the “rioting in the streets” line landed. It is not a prediction I enjoy repeating. It is a reminder that labor shocks are not only labor stories. They become political weather.
A Better Best Case Is Still Available
The best case is not a world without disruption. That ship has sailed. The best case is disruption with ramps. Employers who implement tools without dropping the floor overnight. Governors who rewrite training and insurance instead of issuing another task force photo. Workers who get time, money, and a plausible next role instead of a webinar.
Raimondo still believes the long run can add work, not just subtract it. So do I, with the usual caveats. Technologies have a record of creating tasks that sounded silly right before they became jobs. The catch is brutal and ordinary. People live in the short run. Mortgages live in the short run. Kids’ tuition lives in the short run.
If the country wants the optimistic ending, it has to build the hallway first. Otherwise the new rooms will open and a lot of people will still be standing in the dark, holding a paycheck that no longer exists.
What Readers Should Watch Next
Watch state experiments, not just model releases. Watch whether payroll rules move at all. Watch whether retraining money follows displaced people or follows slogans. Watch youth unemployment as closely as you watch benchmark charts. And watch whether the firms that talk about responsible adoption still treat headcount as the fastest lever in the kit.
The technology will keep improving. That part is not in doubt. The open question is whether the labor system improves on anything like the same clock. If it does not, the optimists will still be right about year fifteen and wrong about the years that decide whether people believe them.
That is the worry hiding inside the hope. Not that machines will think. That institutions will stall. And that by the time everyone agrees the transition needed a plan, the plan will be arriving after the layoffs have already done their work.
