Why People Resist AI (And What Signaling Theory Tells Us About It)

In 2001, George Akerlof won the Nobel Prize for explaining why used-car markets fail. The answer wasn't complicated: when sellers know more than buyers about quality, and anyone can claim their product is good, buyers stop trusting everyone. Good sellers exit. Only lemons remain.

Fifty years later, AI has created the same problem—at scale, across every knowledge profession.

As we enter 2026, the strategic question has shifted from "How do we adopt AI?" to "How do we signal genuine human value in a world where production costs approach zero?"

In 1970, economist George Akerlof asked a deceptively simple question: Why do used car markets fail?

His answer transformed economics. In a market where sellers know more than buyers about product quality, something counterintuitive happens. Good car owners can't prove their cars are good—anyone can claim quality. So buyers assume the worst and offer low prices. At those prices, owners of genuinely good cars refuse to sell. They exit the market. What remains? Only the low-quality vehicles—what Akerlof called "lemons."

Three years later, Michael Spence identified the solution: costly signals. Education, he argued, doesn't necessarily make workers more productive. It works as a hiring signal because it's expensive—in time, money, and effort. Crucially, it's more costly for low-ability individuals to obtain than for high-ability individuals. The cost differential is what makes the signal credible.

In 2001, Akerlof and Spence shared the Nobel Prize in Economics with Joseph Stiglitz for these foundational insights into asymmetric information. Their work explained why markets don't always self-correct—and why effort, credentials, and reputation matter more than we'd like to admit.

The uncomfortable truth: Signals only carry value when they're costly to fake. And AI just made faking them nearly free.

Meta's $2B+ acquisition of Manus AI signals the shift from conversational AI to autonomous agents that complete entire tasks—research, presentations, data analysis—without constant supervision.

This session demonstrates production-tested workflows, including a style-matching presentation method that combines human content direction with AI design execution, delivering polished decks in under 20 minutes. Attendees leave with immediately actionable frameworks plus honest guidance on current limitations and when traditional AI tools remain the better choice.

🎙️ AI Confidential Podcast - Are LLMs Dead?

☕️ AI Tangle - AI Is Breaking Out of the Screen

🔮 AI Lesson - Meta Just Paid $2B for This AI Agent—Here's How to Actually Use It

🎯 The AI Marketing Advantage - AI Readiness, Ad Strategy Controversy & AI Visibility Trends for 2026

💡 AI CIO - Fresh Minds Outsmart the Experts

 📚 AIOS - This is an evolving project. I started with a 14-day free AI email course to get smart on AI. But the next evolution will be a ChatGPT Super-user Course and a course on How to Build AI Agents.

The resistance isn't irrational—it's an evolved response to signals becoming cheap

I've spent two and a half years writing about AI adoption, and I remain convinced it will reshape how we work. But I've been genuinely surprised by the resistance I encounter—not the kind you'd expect.

The existential risk arguments? Those I understand, even when I disagree. Concerns about job displacement? Legitimate and worth addressing seriously.

But that's not what I'm seeing most often. What I encounter is something more challenging to articulate: people who dismiss AI as "hype," who strongly dislike it, who refuse to engage with it despite clear evidence of its capabilities. These aren't Luddites afraid of change. Many are highly accomplished professionals who've adopted every previous technology wave without complaint.

Something more profound is happening. And I think signaling theory explains it.

The data confirms this resistance isn't anecdotal. According to the Pew Research Center's late 2024 survey of more than 5,000 employed U.S. adults, 81% of workers reported that little or none of their work is done with AI—two years after ChatGPT's public release. Seventeen percent said they hadn't even heard of AI being used in their workplace. A late-2024 EY survey found half of executives observed "fatigue or declining enthusiasm" for AI at their companies. Perhaps most striking: an Upwork survey of 2,500 workers found that 77% of those who use generative AI reported that it has increased their workload rather than reduced it.

And then there's the resistance that goes beyond fatigue. I've heard professionals say, with complete sincerity: "I know it's coming and I'm just not going to use it—even if I lose my job."

That's not fear of displacement. That's something else entirely.

Here's the uncomfortable hypothesis: AI outputs are cheap signals, and humans instinctively devalue them.

When a consultant delivers a strategy document, you're not just paying for the words. You're paying for the years of pattern recognition, the hard-won judgment about what to include and exclude, the professional reputation staked on the recommendation. The document is costly to produce—not only in hours but also in the accumulated investment that made those hours valuable.

When AI generates a similar document in thirty seconds, something breaks. Not the output quality—that might be comparable. What breaks is the signal.

Consider Spence's insight about education: the signal works because it's differentially costly. High-ability people find it easier to obtain than low-ability people. AI inverts this entirely. Now, anyone can produce sophisticated-looking output with minimal effort. The cost differential collapses.

And when signals become cheap, markets fail. Good work becomes indistinguishable from mediocre work wrapped in AI polish. Buyers—clients, employers, readers—can't tell who actually has judgment and who's just running prompts. So they discount everything.

The resistance to AI isn't irrational. It's the same instinct that makes you suspicious of a used car priced too low. When something that should be costly becomes cheap, we instinctively distrust it—because historically, that instinct has been correct.

The Deloitte case study makes this dynamic concrete.

In December 2024, Australia's Department of Employment and Workplace Relations paid Deloitte AU$440,000 (approximately $290,000 USD) to conduct an "independent assurance review" of its welfare compliance system. What the government expected was Big Four rigor—the kind of expert analysis that justifies premium consulting rates.

What they received was a 237-page report riddled with fabricated citations, invented court cases, and hallucinated quotes attributed to real experts.

University of Sydney researcher Chris Rudge spotted the problems immediately. One citation referenced a book by a colleague that didn't exist. "I instantaneously knew it was either hallucinated by AI or the world's best-kept secret," Rudge told the Associated Press. He catalogued roughly 20 errors before alerting media and government officials.

Deloitte eventually acknowledged that the report was produced using Azure OpenAI GPT-4o and agreed to refund a portion of the payment. Australian Senator Deborah O'Neill's response was brutal: "Deloitte has a human intelligence problem... Perhaps instead of a big consulting firm, procurers would be better off signing up for a ChatGPT subscription."

That last line captures the signaling collapse perfectly. If a $290,000 consulting report can be replicated by a $20/month subscription, what exactly was the client paying for? The answer should have been judgment, verification, and professional accountability. Instead, Deloitte delivered AI output without the human oversight that justified the premium.

This isn't an isolated incident. It's a preview of what happens when organizations automate production without rethinking how they signal value.

There are three capabilities that AI cannot credibly signal—and these become the new premium in a cheap-signal economy.

Surviving the cheap signal economy requires rethinking how value is demonstrated, not just delivered.

Phase 1: Audit Your Signal Stack

Map every deliverable your organization produces and ask: "What happens if this can be AI-generated in thirty seconds?" Where your value proposition collapses, you've found your vulnerability. Where it strengthens—because human judgment remains essential—you've found your differentiation.

A law firm's research memo faces a signal collapse. The same firm's courtroom presence, client relationship, and strategic judgment do not. A marketing agency's content production faces a signal collapse. Its understanding of brand voice, audience psychology, and campaign timing does not.

Phase 2: Make Human Value Visible

The firms that win will be those that transparently demonstrate where human judgment occurs. Consider the difference:

Opaque: "Here is our strategic recommendation."

Transparent: "AI-generated initial options. Our senior team evaluated against your specific context, rejected three approaches that would have created compliance risk, and refined the recommendation based on our experience with similar organizations. Partner review and sign-off attached."

The second version is more costly to produce—which is precisely what makes it credible.

Phase 3: Redesign Pricing Around Judgment

If clients are paying for production, AI will commoditize you. If they're paying for judgment, verification, and accountability, AI amplifies your value. This requires explicit unbundling:

Hiding AI use instead of demonstrating oversight. Deloitte's problem wasn't that they used AI. The fact that they used AI without human verification justified their rates. Organizations that conceal AI involvement invite the worst assumption: that they delivered commodity output at premium prices. The better approach is to document and advertise your oversight process.

Competing on production speed alone. If your AI advantage is "we can produce this faster," you're racing to the bottom against everyone with the same tools. Speed without judgment is undifferentiated. The strategic position is "we can produce this faster AND tell you which parts to trust."

Assuming credentials still signal capability. Historic indicators of expertise—degrees, titles, years of experience—signal the ability to produce quality work. When AI collapses production costs, these credentials become necessary but insufficient. What matters now is demonstrated judgment: a track record of decisions that produced good outcomes in ambiguous situations.

Overinvesting in AI adoption without signal redesign. Many organizations treat AI implementation as a technical project. Acquire tools, train staff, deploy widely. This misses the strategic point. Without simultaneously redesigning how you signal value to clients and markets, you're automating yourself into commodity status.

The signal-aware organization captures value in three ways:

Premium pricing for judgment. When production becomes commoditized, judgment becomes scarce. Organizations that clearly demonstrate where human oversight occurs can justify premium rates while competitors race to the bottom on production pricing. The keyword is "demonstrate”—the judgment must be visible to capture the premium.

Client retention through trust accumulation. In a market flooded with cheap signals, relationships become more valuable, not less. The advisor who has earned trust through years of good judgment becomes irreplaceable precisely because AI makes initial engagement easier and ongoing trust harder. Client relationships are moats that AI cannot cross.

Talent attraction and retention. Knowledge workers increasingly resist organizations that treat them as prompt operators. Firms that position AI as amplifying human judgment—rather than replacing it—attract professionals who want to exercise expertise, not supervise machines. This is a recruiting advantage that compounds over time.

Quantifying the opportunity: The Deloitte case involved a $290,000 contract. The reputational damage likely exceeds that by an order of magnitude. Organizations that avoid signal collapse protect both immediate revenue and accumulated brand value. For professional services firms, this represents millions in preserved trust equity.

The resistance to AI isn't going away—not because people are irrational, but because they're correctly perceiving a genuine problem. Cheap signals devalue markets. Humans have evolved to distrust them. No amount of AI enthusiasm will overcome this instinct with persuasion alone.

The path forward is to restore the credibility of human value. This means transparency about how AI is used, explicit demonstration of where human judgment occurs, and pricing structures that separate commodity production from premium oversight.

The strategic question isn't whether to adopt AI. It's how to maintain signal integrity when production costs approach zero. Organizations that solve this problem will command premium positioning. Those that don't will find themselves in Deloitte's position: unable to justify their rates, unable to distinguish themselves from a $20 subscription.

The used car market fixed itself through warranties, inspections, and reputation systems—all mechanisms to restore signal credibility. Knowledge work needs equivalent innovations. The firms building those innovations now will define the next era of professional services.

What's your warranty?

Author’s note: This week’s complete edition—including the AI Toolbox and a hands-on Productivity Prompt—is now live on our website. Read it here.

Join us at All Things AI 2026 in Durham, North Carolina, on March 23–24, 2026.

This is where AI gets real. No sales pitches—just 4,000 builders, engineers, operators, and execs sharing how AI actually works in practice, from hands-on workshops to real-world sessions, right in the heart of RTP and Duke’s AI ecosystem.

Early registration is open, and prices go up soon.

I appreciate your support.