Reflections on "The Economics of Bicycles for the Mind" - The Future of Expertise and AI

Understanding the Research: From Metaphor to Economic Theory

Steve Jobs once famously described computers as "bicycles for the mind" - tools that dramatically leverage human capabilities rather than replace them. In their July 2025 NBER working paper, "The Economics of Bicycles for the Mind", Agrawal, Gans, and Goldfarb transform this metaphor into rigorous economic theory, presenting a formal model of how cognitive tools and technologies enhance mental capabilities.

The paper examines AI agents engaged in iterative task improvement and establishes critical distinctions in how cognitive tools interact with different types of human skills. Cognitive tools are assumed to be substitutes for implementation skills—the technical execution aspects of tasks. When a tool can handle data processing, computation, or pattern recognition, it directly replaces the need for humans to perform these implementation functions.

But the relationship with judgment proves more nuanced. The authors identify two distinct types:

• Opportunity Judgment: The ability to recognize opportunities to start or improve a process - shown to always complement cognitive tools

• Payoff Judgment: The ability to know which action to take in a given state—not necessarily a complement to cognitive tools

This framework synthesizes seemingly contradictory empirical evidence. While both computers and AI increase productivity, their impacts on inequality diverge: computers have contributed to increased inequality, while AI shows both increases and decreases depending on context. The model explains these differences through the varying relationships between cognitive tools and types of judgment.

Validation: the Human × AI Strategic Opportunity

This groundbreaking research validates what forward-thinking organizations are discovering: in complex, high-stakes situations, the future belongs not to automation but to Human × AI collaboration where judgment, not technical skill, becomes the critical edge. The economists who literally wrote and edited the foundational books on AI economics (Prediction Machines, Power and Prediction, The Economics of Artificial Intelligence) have provided rigorous theoretical foundation for why this approach creates sustainable competitive advantage.

The Three-Layer Model of Human Capability

The paper places human skills into three distinct categories:

Implementation Skills: Technical execution and process management. This is where cognitive tools excel—handling data processing, pattern recognition, and computational complexity at machine speed.

Opportunity Judgment: Identifying improvement opportunities and strategic openings. This requires contextual awareness and creative insight that pre-programmed systems cannot provide.

Payoff Judgment: Translating insights into optimal decisions and actions. This is where experience, wisdom, and strategic thinking create irreplaceable value.

The paper's core finding: cognitive tools consistently substitute for implementation skills while complementing judgment capabilities. Productivity gains come from human-AI collaboration rather than full automation, with tools amplifying human cognitive abilities rather than replacing them.

Why Automation Hits a Wall

The research reveals a fundamental limitation: automation requires "pre-specified judgment." You can only automate what you can fully anticipate. But complex tasks demand adaptive, contextual decision-making that only humans can provide.

Consider what happens in a high-stakes merger negotiation when a regulatory body unexpectedly shifts position, a key stakeholder changes priorities, or new information surfaces that reframes the entire deal structure. No pre-programmed system can navigate these moments. They require human judgment to recognize the shift, assess implications, and adapt strategy—all while maintaining relationships and managing multiple stakeholder dynamics.

The paper also reveals a U-shaped inequality curve: AI initially reduces skill premiums by substituting for implementation capabilities, but eventually increases inequality as judgment becomes more valuable relative to technical execution. As cognitive tools handle more implementation tasks, the scarce resource becomes the ability to direct that execution wisely.

The New Architecture of Expertise

The NBER framework reveals how ALL professional services must fundamentally restructure. The traditional model—senior experts supervising junior staff handling implementation—collapses when AI handles implementation better, faster, and cheaper than any human. The emerging architecture requires reorganization around judgment layers rather than seniority hierarchies.

Law firms that once deployed armies of associates for document review now need senior partners who can recognize which patterns matter. Consulting firms that built models with analyst teams now require leaders who know which questions to ask. Investment banks that processed deals through rigid hierarchies must develop professionals who combine opportunity recognition with tactical wisdom.

Take complex negotiations as a concrete example. At Expeditionary, we're architecting our collaborative AI systems around this new reality to help navigate complex negotiation:

AI Implementation Layer: Our platform handles data processing, stakeholder mapping, scenario modeling, and pattern recognition - the implementation skills where cognitive tools excel.

Human Judgment Layer: Our negotiation experts provide strategic opportunity identification, cultural intelligence, tactical decision-making, and real-time adaptation—the judgment capabilities that cognitive tools complement rather than replace.

This isn't about adding AI to existing processes. It's about fundamentally restructuring how expertise creates value. Our Human × AI™ negotiation platform demonstrates what happens when you design from first principles around the complementarity between cognitive tools and human judgment.

The Learning Laboratory Effect

There's an overlooked benefit to engaging with Human × AI systems in high-stakes situations: it serves as a learning laboratory for broader organizational transformation.

Leaders who experience firsthand how AI amplifies judgment while handling implementation gain direct understanding of how to restructure their own organizations. They see which decisions require human judgment, where AI excels, and crucially, how to orchestrate the interaction between them. This experiential learning—under real pressure with real stakes—provides insights no workshop or whitepaper can deliver.

When executives work alongside our Human × AI negotiation platform, they're not just achieving better deal outcomes. They're experiencing the future of expertise in action. They observe how our negotiators direct AI's analytical power while maintaining relationship dynamics. They see how opportunity judgment identifies moments others miss. They understand viscerally why payoff judgment—knowing which action to take—remains irreplaceably human.

This experience transforms how leaders think about their own organizations. Which roles exist primarily for implementation? Where does judgment create differential value? How should teams restructure when cognitive tools handle elements of execution? These questions move from abstract to concrete when you've lived through the transformation in a high-stakes context like a complex negotiation process.

The Competitive Reality: Why This Matters Now

We're at an inflection point. The research demonstrates that firms with superior Human × AI integration will gain increasing competitive advantages over those relying on either pure human approaches or full automation strategies.

For C-suite executives, this creates urgency: if you're not enhancing your capabilities with Human × AI collaboration, your competitors will be. Even modest improvements of 5% create substantial competitive advantages—on a $10 billion M&A transaction, that represents $500 million in value creation.

The economics are clear: as cognitive tools become more sophisticated, judgment skills become increasingly valuable, creating premium opportunities for firms that can effectively integrate human expertise with AI capabilities.

The Emerging Executive Playbook: Expertise and AI

Leaders navigating this transformation need to answer critical questions:

Where to Preserve Human Judgment vs. Deploy AI Implementation

• Map your organization's work across the three categories: implementation, opportunity judgment, and payoff judgment

• Identify where pre-specified rules work (automate) vs. where adaptive judgment is essential (augment)

• Recognize that the highest-stakes decisions will always require human judgment

How to Identify Which Expertise Becomes More vs. Less Valuable

• Technical specialists who primarily execute will see their roles transformed

• Professionals who recognize patterns, identify opportunities, and exercise judgment become more valuable

• The ability to direct and orchestrate AI effectively becomes a core competency

Questions Every CEO Should Ask

• Which of our current processes assume human implementation that AI could handle better?

• Where are we using senior talent for tasks that are actually implementation masked as judgment?

• How do we restructure teams when the junior-to-senior progression no longer makes sense?

• What new capabilities emerge and what new value can be created when our people are freed from certain implementation tasks?

The Path Forward

The paper's finding that automation faces inherent limitations in judgment-intensive tasks points to a clear future: organizations that thrive won't be those with the most powerful AI or the most experienced professionals in isolation—they'll be those who best integrate both into genuine collaboration and co-intelligence.

Steve Jobs' bicycle metaphor is becoming economic reality. Cognitive tools don't replace minds; they amplify them. But only for those who understand the new architecture of expertise and restructure accordingly. At Expeditionary, we're building this future. When it comes to complex negotiation, attempts at AI automation will faces inherent limitations in judgment-intensive tasks. Complex negotiations require adaptive strategy, cultural intelligence, and real-time tactical adjustments that pre-programmed systems cannot provide. It's not a question of keeping humans in the loop - humans are the loop. That's why we're developing collaborative AI solutions that are purpose-built for complex negotiation and combining that with our decades of experience pioneering the application of negotiation foundational methodologies in real-world, high-stakes contexts.

We believe humanity's most important negotiations deserve our most powerful capabilities. When billions of dollars, war and peace, critical relationships, and strategic futures are on the table, you need the multiplication of capabilities that creates outcomes neither human nor AI could achieve separately.

The economics are proven. The technology exists. Those who recognize that amplifying human expertise and judgment creates competitive edge will define the decades ahead.