CitriniResearch and Alap Shah’s 2/22/26 piece titled “The 2028 Global Intelligence Crisis” is thought-provoking, well-researched, and beautifully written. It lays out a scenario of economic and societal collapse driven by the unhinged success of artificial intelligence, and how it could significantly alter the structure of our workforce and economy. If the authors' goal was to tell an entertaining story, publish an article that could go viral, and/or scare many investors, I think they succeeded on all fronts.
But as I was reading the article, I couldn’t help but think about how the evolution of our country over the past 250 years was due to ingenuity, innovation, hard work, intelligence, and strong leadership. And, how capitalism has always prevailed. So, I thought, what if Citrini and Shah have it backwards? What could the inverse view, the positive perspective, look like?
I couldn’t get it out of my head, so at 10:04 PM last night, I copied the report and pasted it into ChatGPT. Then I added this prompt: “Review the article and write a similar-length article making a case for the other side. Make a case for how AI will help the economy for decades, how workers will become more productive, and how the US will thrive.” In less than ten seconds, I was reading a very different yet compelling story.
While the future may lie somewhere between the extremes, personally, I choose optimism over pessimism.
Enjoy!
(The output from ChatGPT has not been edited except for removing a few proper names.)
Preface
What if our AI bullishness continues to be right… and what if that’s actually bullish?
What follows is a scenario, not a prediction. Not “AI solves everything” fan-fiction, and not a hand-wave over real dislocation. The point is to model a path that’s been underweighted in many macro takes: the one where AI is a general purpose technology that raises the long-run human premium by expanding the frontier of what humans can do, the way electrification, computing, and the internet did—just faster and messier at first.
Think of this as the counter-memo: a reconstruction of how the U.S. economy could look better—not because AI is gentle, but because the U.S. is unusually good at turning productivity shocks into new industries, new jobs, and new consumer demand.
This is a memo from June 2028, detailing the progression and upside of the Productive Abundance Cycle.
The Consequences of Augmented Intelligence
February 22nd, 2026 → June 30th, 2028
The unemployment rate printed 4.1% this morning, a 0.2% downside surprise. The market rallied 1.6% on the number, bringing the cumulative gain in the S&P to 34% from its October 2026 highs.
Two years. That’s all it took to get from “agentic AI is a toy” to an economy that no longer resembles the one any of us grew up in.
But here’s the twist: it got weirder in exactly the way long-run growth people have been begging for since 2008. Not “ghost GDP.” Not a hollow boom that never touches households. Something else—an acceleration in capabilities diffusion that widened access to expertise, lowered the effective cost of starting and scaling a business, and made the median worker meaningfully more productive.
This quarter’s memo is our attempt to reconstruct the sequence.
The headline numbers stayed good… and then got better
Nominal GDP repeatedly printed mid-to-high single-digit annualized growth. Productivity surged, yes—but this time the benefits didn’t stall in the balance sheet of “owners of compute.”
Real output per hour rose at rates not seen since the post-war era. But unlike the “automation scares” of prior decades, the wage data eventually caught up. Not immediately. Not evenly. But measurably. Real compensation for complementary roles accelerated, and the labor market re-sorted around what humans do best when intelligence is cheap: judgment, taste, relationships, accountability, and complex multi-stakeholder coordination.
The early debate framed AI as a labor-replacing machine. In practice, it acted more like a force multiplier—a layer of universal assistants that expanded what one person could supervise, create, diagnose, negotiate, and ship.
The core macro result was simple: when the cost of cognition falls, demand for outcomes rises. And demand—especially in a consumption-heavy economy—has a way of finding the new capacity.
How It Started: In late 2025, agentic tools crossed an adoption threshold
The story begins where most stories begin now: software.
A competent developer working with next-gen coding agents could replicate the core functionality of a mid-market SaaS product in weeks. Investors looked at that and saw a pricing collapse. They weren’t wrong about the pricing pressure. They were wrong about the conclusion.
By mid-2026, procurement teams did what procurement teams do: they demanded discounts. The long tail of SaaS—workflow tools, task trackers, “glue” software—felt the squeeze first.
But then something more interesting happened.
SaaS didn’t die. It unbundled.
- Incumbents dropped prices on commoditized features.
- They raised prices on integrated workflows, compliance, security, governance, and uptime guarantees.
- The “seat” model shifted toward “outcome” pricing.
- The biggest beneficiaries weren’t the vendors. It was the buyers—who reinvested the savings into expansion, hiring, and new projects that had been uneconomic before.
The mistake in the bearish model was assuming cost savings equal demand destruction. In reality, cost savings are often demand creation—especially when they arrive as broad-based reductions in friction across thousands of businesses at once.
The first wave wasn’t layoffs. It was re-composition
Yes, some headcount fell in certain white-collar functions—especially repetitive drafting, basic research, entry-level coding, and routine customer support.
But the dominant corporate response wasn’t “replace humans with machines.” It was “raise the output target.”
Boards didn’t look at AI and say, “great, we can shrink.” They said, “great, now we can finally ship what we’ve been promising.”
Companies used AI to:
- shorten product cycles,
- expand customer coverage,
- reduce defect rates,
- speed up onboarding,
- improve sales enablement,
- and compress time-to-revenue.
What looked like “labor substitution” in a spreadsheet became “capacity expansion” in the real economy.
When Friction Went to Zero: The consumer economy didn’t wither. It got sharper
By early 2027, LLM usage had become default. People used AI agents the way they used maps: not as a novelty, but as infrastructure.
The bearish view was: if agents optimize purchases, margins collapse, intermediation dies, and the consumer economy shrinks.
The reality was subtler: agents reorganized competition.
Yes:
- subscription renewals became harder to trap,
- junk fees got pressured,
- lazy price discrimination eroded,
- and some toll-booth models broke.
But those weren’t “the economy.” Those were rents.
When you remove rents, you don’t remove spending—you redirect it. Households didn’t stop consuming; they started consuming more efficiently, with higher satisfaction per dollar. The net effect looked a lot like a broad-based tax cut, except delivered via technology.
The new consumer surplus showed up quickly in categories where people buy outcomes, not brands:
- travel experiences (more trips, better planning, less waste),
- home improvement (projects scoped correctly the first time),
- health and wellness (better adherence, more preventative care),
- education (cheap tutoring at scale),
- and small luxuries (because people could finally afford them again).
Agents made the economy less extractive and more responsive. That’s deflationary in some line items, but not in a depressionary sense. It’s the good kind of deflation: lower costs, higher real living standards, rising real demand.
Intermediation didn’t vanish. It evolved
The bearish narrative treats intermediaries as pure friction. Many were. Some deserved to be disrupted.
But the durable intermediaries didn’t sell “navigation of tedium.” They sold:
- trust,
- liability,
- curated risk,
- compliance,
- and human accountability when the stakes are high.
AI shifted the stack from “human does everything” to “AI does the drafting and analysis; humans do the assurance.”
This was clearest in regulated domains:
- healthcare (clinical decision support plus clinician accountability),
- finance (automated analysis plus fiduciary responsibility),
- law (automated research plus legal judgment),
- and enterprise security (automated monitoring plus governance).
In each, AI reduced the labor hours per unit of outcome—then demand expanded because the outcomes got cheaper and more accessible.
From Sector Risk to Systemic Upside: The U.S. advantage wasn’t chips. It was diffusion
Taiwan and Korea did great selling picks and shovels. The U.S. did something else: it spread capabilities into millions of firms and households faster than anyone else.
That’s always been the American superpower.
The U.S. has:
- massive domestic demand,
- deep capital markets,
- high labor mobility,
- a culture of entrepreneurship,
- and an unmatched ability to commercialize software.
When intelligence became abundant, the binding constraint stopped being “can we figure it out?” and became “can we execute?” That shift favored a country that is unusually good at execution at scale.
The “new job” question answered itself—through new firms
The ATMs analogy missed the point. The internet analogy got closer. But AI’s real pattern looked like cloud + smartphones + open source all at once.
The biggest job engine wasn’t AI labs. It was the explosion of small and mid-sized businesses that could finally operate like large enterprises.
In 2027–2028 we observed a structural change:
- A two-person startup could run marketing, finance, customer support, legal drafting, product iteration, and analytics at a level that previously required 15–30 people or expensive vendors.
- “Solo founders” became viable in more categories.
- Local businesses became more sophisticated: better pricing, better inventory management, better customer retention.
The old world had a steep fixed cost of competence. The new world flattened it.
That matters because the U.S. doesn’t just create jobs by scaling incumbents. It creates jobs by birthing new firms—then competing them into entire new sectors.
The Productivity Dividend: The biggest second-order effect was time
The bearish model asks: “how much money do machines spend?”
The bullish model asks: “what do humans do with the time they get back?”
Across the economy, AI created a new kind of resource: reclaimed hours.
Reclaimed hours translated into:
- more output per worker,
- more entrepreneurship,
- more caregiving capacity,
- more education and reskilling,
- more second jobs and side businesses (often creative or local),
- and, for many, better quality of life.
This matters macro-economically because time is upstream of both productivity and consumption. When you increase the efficiency of the “planning and coordination layer” of life, you increase the feasible set of what people will attempt.
Wages: the human premium didn’t die, it polarized—then broadened
There was a painful transition in entry-level white-collar roles that used to be “apprenticeship through busywork.” That pipeline broke and had to be rebuilt.
But the human premium didn’t disappear. It moved.
It concentrated in:
- domain expertise,
- taste and communication,
- ownership and accountability,
- leadership in ambiguous environments,
- and embodied work that benefits from AI guidance (healthcare, trades, field service, advanced manufacturing).
By 2028, the premium began to broaden again as training improved:
- AI tutors and apprenticeships rebuilt entry pathways,
- firms redesigned jobs around human strengths,
- and workers learned to treat AI like power tools rather than competition.
The average worker became more productive, and in a tight labor market, productivity gains tend to be shared.
The Financial System Didn’t Crack—It Repriced and Adapted: Credit quality held up better than feared
The “mortgage question” turned out to be less dramatic because the labor market never entered a structural freefall. Income volatility rose in some sectors, but default rates remained contained.
Why?
Because two stabilizers kicked in: Policy moved earlier than expected:
- Rapid redeployment: workers moved faster into complementary roles, often with AI making them effective sooner.
- Policy moved earlier than expected: not as a grand “UBI,” but as targeted training credits, wage insurance pilots, and incentives for apprenticeship programs.
The private markets story was also different than doomers expected. Many software LBOs were over-levered, yes—but the broader economy’s strength gave lenders time to restructure rather than liquidate. The losses were real, but they were absorbed in a growing nominal GDP environment.
Equity markets learned to value diffusion, not just compute
In 2026 the market was “AI equals chips.” By 2028 the market was “AI equals everything that adopts.”
The winners expanded from:
- GPU makers and hyperscalers to include:
- industrial automation,
- logistics,
- healthcare services,
- education platforms,
- cybersecurity,
- and millions of “boring” firms that quietly doubled output per employee.
The AI trade stopped being narrow and became broad—exactly what you want if you’re looking for durable, economy-wide growth rather than a top-heavy bubble.
The Role of Policy: The government didn’t need to own AI—just to accelerate mobility
The most effective interventions were the least ideological:
- portable benefits,
- faster credentialing,
- apprenticeships with wage subsidies,
- modernized unemployment insurance tied to training,
- and tax incentives for companies that proved they were upskilling workers rather than simply shedding them.
The most important shift wasn’t redistribution. It was re-tooling the labor market for speed.
Institutions adapted more quickly than expected because the incentives aligned. Firms needed talent. Workers needed pathways. Politicians needed a story that wasn’t “the future is unemployment.”
The story became: the future is leverage.
The Abundance Flywheel
By mid-2028, we could finally name the mechanism:
AI improved → costs fell → outcomes got cheaper → demand rose → firms expanded → workers became more productive → wages rose in complementary roles → consumption strengthened → more investment flowed into diffusion → AI improved.
A flywheel, not a spiral.
The biggest misunderstanding in the bearish scenario was treating the economy like a fixed pie of tasks. In reality, it’s a dynamic system of wants. When you make it cheaper to build, coordinate, diagnose, design, and deliver, you don’t just replace labor—you expand the set of things society is willing to pay for.
And Americans, as always, found new things to want:
- better healthcare,
- better homes,
- better experiences,
- better education,
- better security,
- and more personalized everything.
AI didn’t reduce the economy. It raised the ceiling.
The Human Premium Reframed
For modern economic history, human intelligence was scarce and expensive. We built institutions around that scarcity.
Now intelligence is abundant—but humanity is still scarce.
- Trust is scarce.
- Taste is scarce.
- Accountability is scarce.
- Courage in ambiguity is scarce.
- Care is scarce.
- Leadership is scarce.
AI commoditized certain cognitive tasks. It didn’t commoditize being a person responsible for outcomes in a messy world.
The premium didn’t vanish. It moved from “can you produce information?” to “can you produce decisions, relationships, and results?”
That is a different labor market. Not necessarily a worse one.
Closing
You’re not reading this in June 2028. You’re reading it in February 2026.
The S&P is near all-time highs. The debates are loud. The demos are intoxicating. The risks are real. The transition will be uneven, and some business models built on friction will die.
But the base case should not be economic collapse. The base case—if diffusion continues, if institutions adapt, and if the U.S. leans into its strengths—is that AI becomes the productivity engine we’ve been missing for decades.
The canary is still alive.
And it might be learning to sing in harmony with the machine.
The content in this blog was AI generated, unedited, based on a prompt that requested a specific answer in the output, and does not reflect the views of Ullmann Wealth Partners. Private investments involve additional risks, including limited liquidity, and may not be suitable for all investors. Past performance is not indicative of future results.