The promise was seductive: AI would be your colleague, not your replacement. It would handle the drudgery—the repetitive emails, the tedious data entry, the endless formatting chores—while you ascended to more meaningful work. For years, that’s the pitch that sold enterprise AI to organizations hungry for productivity gains. But something strange is happening on the ground. The technology that was supposed to liberate workers is instead trapping them in a new kind of treadmill—one that feels faster, more intense, and paradoxically less fulfilling than what came before.
A Wall Street Journal analysis captured this paradox in stark terms: AI is increasing the speed, density, and complexity of work rather than reducing it. The very tool marketed as an antidote to burnout is becoming a catalyst for it. To understand why, we need to look not at the algorithms themselves, but at the human brains operating them—and the psychological forces that make productivity tools feel more like dependencies.
The Dopamine Economy of AI Work
The mechanism at play isn’t primarily about poor implementation or insufficient training. It’s rooted in the same neurological circuitry that makes social media so difficult to resist. When you use AI to complete a task—whether it’s drafting a response, generating a slide, or summarizing a document—you receive an immediate hit of satisfaction. Your brain registers a small victory. Dopamine floods the reward pathway, and you feel productive, capable, engaged.
The problem is that this feeling is inherently fleeting. Just as a single social media like provides only momentary gratification, a single AI-assisted task leaves you craving the next dose. Peter Cowen, Forbes senior contributor and author of Brain Rush, draws a direct parallel: “It’s the same kind of dopamine addiction that has happened with social media.” The cycle becomes self-reinforcing. You complete a task, you feel good, you reach for the next one. The technology that was supposed to create breathing room instead becomes a machine for filling every available gap.
This creates what might be called the productivity paradox: the more efficient you become, the more efficient you feel compelled to be. Workers aren’t necessarily being forced to work longer hours by explicit employer demands. Instead, they’re trapped by an internal drive—a psychological need to keep achieving, to keep producing, to keep capturing that dopamine hit. The fear is real, even if the threat is often imagined. As Cowen observes, many employees operate under a persistent anxiety: “If they don’t spend a lot of time working and using their AI, they might be just fired or replaced.” So they fill every void. They answer emails at midnight. They run queries before breakfast. They never stop, because stopping feels like falling behind.
The Crowding Out of Creativity
The data paints a troubling picture. Workers are spending more than double the amount of time on email and messaging compared to pre-AI baselines. Meanwhile, time spent on what researchers describe as “focused uninterrupted work”—the kind of deep, creative, strategic thinking that actually moves careers and organizations forward—has fallen by 9%.
This isn’t a coincidence. It’s a crowding-out effect. The immediate reward loop of AI-assisted task completion makes creative work feel inefficient by comparison. Creative work rarely delivers instant gratification. It requires periods of incubation, ambiguity, and uncertainty before results emerge. You might spend an hour on a problem and emerge with nothing. You might need to go for a run, take a shower, let your mind wander before the insight arrives. That’s precisely what makes creative work valuable—but it’s also precisely what makes it feel unrewarding in the context of an AI-accelerated workflow.
Cowen describes this vividly: “When I do the best creative work, it’s sort of when I’m not thinking about it. If I’m running, I will have a problem that’s sort of stewing in the back of my mind, and then I won’t even know the answer, but then I’ll just be running and all of a sudden the answer will pop into my mind.” This is the experience that’s being displaced. The constant pinging, the steady stream of AI-assisted completions, the feeling of being always “on”—these create a kind of noise that drowns out the quiet necessary for genuine innovation.
The Value Pyramid: Where Most Companies Get Stuck
Cowen proposes a framework that helps explain why AI implementation so often fails to deliver on the promise. He calls it the “value pyramid”—a three-tier model that maps how organizations actually use artificial intelligence.
At the base of the pyramid sits what Cowen calls “overcoming creator’s block.” This is the most common use case: employees using AI to get started on emails, to generate first drafts of presentations, to overcome the initial paralysis of a blank page. It’s useful, but it’s fundamentally defensive—it’s about getting through the day, not about moving the organization forward.
The middle tier involves “getting more productive”—using AI to do more work in less time, to compress processes, to squeeze greater output from existing resources. This is where most companies aspire to be, and it’s where the dopamine trap is most pronounced. Productivity increases, but the gains are captured by the organization rather than translated into leisure or well-being for workers.
The top of the pyramid—where only a small minority of companies operate—involves “creating new growth curves.” This is the transformative use of AI: identifying unmet social needs, building products that address them, using AI as an enabler of entirely new business models. It’s the difference between using AI to do things faster and using AI to do different things entirely. The key insight here is that this level of implementation requires something most organizations lack: entrepreneurial leadership that can articulate a vision for what AI enables, not just distribute the tools and hope for the best.
The ROI Problem: Why Most AI Investments Aren’t Paying Off
The numbers are striking: approximately 95% of companies piloting AI are not seeing a return on their investment. Only 5% are achieving meaningful ROI. This isn’t simply a matter of implementation challenges—it’s a structural issue rooted in how organizations approach the technology.
The hyperscalers—Amazon, Google, Microsoft—are spending hundreds of billions on AI infrastructure and seeing returns through cloud growth and advertising optimization. But for the typical enterprise, the economics are murkier. The cost of running AI models, particularly the most sophisticated ones, can be enormous. Some companies are beginning to find efficiencies by matching computational tasks to appropriate model tiers—using cheaper models for simpler jobs and reserving expensive models for complex analysis—but this requires a level of strategic sophistication that most organizations haven’t developed.
The deeper problem is that most CEOs don’t know how to lead the kind of transformation that would justify these investments. Cowen observes that “most CEOs do not know how to lead a business process change that will allow a company to have some advantage over the competition.” They’re giving employees access to AI and letting them figure it out—a recipe for the dopamine-driven hamster wheel rather than strategic advantage.
The 3-Day Work Week Myth
Perhaps no promise of the AI revolution has been more widely circulated—and more widely doubted—than the suggestion that workers might soon enjoy a three-day or four-day work week. Bill Gates, Jamie Dimon, and other prominent CEOs have suggested that AI could make this a reality. The logic seems sound: if machines handle more work, humans can work less.
But Cowen’s assessment is blunt: this is unlikely to materialize in any way that benefits workers. “If you have people who can work 3 days a week, why would a company employ that person for only 3 days a week?” he asks. The more probable outcome is that companies will simply require existing employees to produce more output, reducing headcount rather than hours. The productivity gains won’t be distributed as leisure; they’ll be captured as cost savings and distributed to shareholders.
This reflects a fundamental misalignment between the promises made by executives and the incentives embedded in corporate structures. A CEO might genuinely believe that AI will create a world where humans work less—but that belief doesn’t change the fact that the organizations they lead are designed to maximize output per dollar of labor cost. If AI makes workers more productive, the rational corporate response is to employ fewer of them, not to give the existing ones more free time.
The Path Forward: Reclaiming Human Well-Being
The challenge isn’t to abandon AI—that’s neither realistic nor desirable. The challenge is to deploy it in ways that serve human flourishing rather than exploiting human psychology. This requires changes at multiple levels.
At the individual level, workers need to become aware of the dopamine dynamics at play. Not every task deserves AI acceleration. Some work is better done slowly, with space for reflection and creativity. The goal isn’t to use AI on everything; it’s to use it strategically on the right things.
At the organizational level, leaders need to move beyond the value pyramid’s base and middle tiers. This means developing genuine strategic visions for how AI enables new forms of value creation—not just using it to make existing processes faster. It also means rethinking how productivity is measured. If output per hour is the only metric, AI will always be used to compress more work into less time. But if the metric includes employee well-being, creative output, and sustainable performance, the calculus shifts.
At the societal level, we need a broader conversation about what work is for. The assumption that productivity growth should translate entirely into more output—that leisure is a residual rather than a right—reflects a set of choices that could be made differently. The technology doesn’t determine the outcome; the institutions and incentives surrounding it do.
The irony is stark: we’ve built tools meant to reduce labor and increase leisure, and somehow ended up with more intensity, more pressure, and less space for the human qualities that make work meaningful. The technology isn’t the villain. But the way we’re deploying it—without regard for psychological impact, without strategic vision, without concern for well-being—may be costing us exactly what we hoped to gain.