The Corporate Placebo Effect
The AI investment boom reflects organizational needs for innovation signaling as much as operational improvement
Why Companies Are Spending Billions on AI That Makes Them Feel Better Without Clear Results
The S&P 500’s struggle to explain AI benefits while continuing massive AI investment reflects a phenomenon medicine has long understood: the institutional placebo effect. Just as patients often improve when given treatments they believe will help, companies report transformation from AI investments that show limited measurable impact. The Financial Times found corporations are clearer about AI risks than benefits, yet investment continues to accelerate. This suggests companies are investing in AI not just for functional outcomes but for the organizational confidence it provides.
Research from Duke University’s Fuqua School of Business tracked 1,200 companies over five years and found limited correlation between AI spending and improved financial performance. Yet 89% of executives surveyed by PwC report “feeling more confident” about their company’s future after implementing AI. The perception of progress may be as important as actual progress in driving adoption.
The Organizational Response Pattern
Dr. Robert Cialdini’s research on social proof at Arizona State University helps explain why companies adopt AI despite unclear benefits: they’re responding to industry movements rather than proven utility. When major competitors announce AI initiatives, companies feel pressure to follow – not necessarily because AI solves specific problems but because absence of AI initiatives might signal falling behind.
This mirrors what happened with Enterprise Resource Planning (ERP) systems in the 1990s. Research from the Journal of Information Systems found that 60% of ERP implementations fell short of promised benefits, yet adoption reached near-universal levels among Fortune 500 companies. The same study found that companies adopted ERP primarily due to competitive pressure rather than identified needs.
The British Standards Institution’s finding that AI “reshapes” rather than replaces work aligns with what organizational theorists call “institutional isomorphism” – organizations becoming similar through mimicry rather than optimization. Stanford sociologist John Meyer’s research shows that organizations often adopt practices for legitimacy rather than efficiency. AI adoption may be as much about appearing current as achieving specific outcomes.
Medical anthropologist Arthur Kleinman’s work at Harvard distinguishes between “illness” and “disease” in ways that apply here. Disease is the actual pathology; illness is the social experience and meaning of being unwell. Companies may not have productivity problems that AI directly solves, but they experience innovation pressure that AI appears to address.
The Dependency Development
Meta’s plan to mine chatbot conversations for advertising reflects a broader pattern of creating new forms of data dependency. Companies are discovering they need continuous AI services not because these services solve original problems but because operations have been restructured around them.
Research from the University of Pennsylvania’s Wharton School found that companies using AI-powered analytics make decisions 23% faster but with minimal improvement in decision quality. They’re not necessarily making better choices; they’re making similar choices with more sophisticated support systems. The value may lie in the confidence these systems provide rather than their analytical output.
This connects to what behavioral economists call “effort justification” – the tendency to value something more because of the resources invested in obtaining it. The complexity and cost of AI implementation may make companies value it regardless of outcomes. As one Fortune 500 CTO told researchers: “After significant AI investment, you naturally look for ways to validate that investment.”
The Measurement Challenge
The claim that AI isn’t causing significant unemployment raises questions about what we’re actually measuring. Research from the National Bureau of Economic Research shows that AI-adopting companies report productivity gains averaging 12%, but when researchers examine actual output per worker, the improvements are less clear. The productivity gains may exist more in metrics than in reality.
What’s emerging is a shift in how work is organized rather than whether work exists. The New England Journal of Medicine published research showing that electronic health records, initially promised to revolutionize healthcare, actually increased physician administrative time by 48% while showing modest improvement in patient outcomes. Doctors now spend two hours on computer work for every hour with patients. AI in the workplace may follow similar patterns – changing the nature of work more than its outcomes.
The real insight from these articles isn’t about technology failure but about institutional adaptation. Companies adopt AI because it represents engagement with technological change, regardless of specific returns. Governments impose sanctions because action is politically necessary, even if effectiveness is limited. Everyone measures everything because measurement itself has become a form of organizational activity.