Gary Shapiro has spent more than four decades watching technology reshape industries. As the CEO of the Consumer Technology Association and the man who runs CES, the world’s largest tech trade show, he’s had a front-row seat to every major digital transformation since the personal computer. So when Shapiro sounds an alarm, the C-suite tends to listen.
His latest warning is deceptively simple: artificial intelligence is making leaders dumber.
Not because the technology itself is flawed. But because the way executives are deploying it — as an oracle rather than a tool — is eroding the independent thinking that separates competent management from great leadership. Writing in Fortune, Shapiro argues that the rush to integrate AI into every layer of corporate decision-making has created a dangerous feedback loop, one where leaders increasingly defer to algorithmic recommendations and, in doing so, converge on identical strategies. The result isn’t efficiency. It’s groupthink at scale.
“When everyone uses the same AI tools and gets the same answers, you don’t get better decisions — you get the same decision, everywhere,” Shapiro wrote. The observation cuts against the prevailing narrative in Silicon Valley and on Wall Street, where AI adoption is treated almost universally as a competitive advantage. Shapiro isn’t arguing against adoption. He’s arguing against abdication.
The distinction matters enormously right now. Corporations are pouring unprecedented capital into AI systems. Global spending on AI infrastructure and services is expected to exceed $300 billion in 2026, according to IDC estimates. Every Fortune 500 company has an AI strategy. Most have appointed chief AI officers. And the pressure to show returns on those investments is pushing executives to rely on AI outputs for everything from pricing decisions to workforce planning to M&A target identification.
But here’s the problem Shapiro identifies: when the same large language models and enterprise AI platforms power decisions across competing firms, strategic differentiation collapses. If Company A and Company B feed similar market data into similar models and receive similar recommendations, the competitive moat disappears. The companies don’t outthink each other. They mirror each other.
This isn’t theoretical. Evidence is already emerging across multiple industries. In financial services, quantitative trading firms using similar AI models have experienced increasingly correlated trading patterns, amplifying market volatility during stress events. The Bank for International Settlements flagged this risk in a 2024 report, noting that “herding behavior” among AI-driven trading systems could exacerbate systemic risk. In retail, companies relying on the same demand-forecasting AI tools have found themselves launching near-identical promotions at the same time, cannibalizing each other’s margins without gaining market share.
And in hiring, the pattern is even more stark. Multiple studies have shown that AI recruitment tools trained on similar datasets produce remarkably homogeneous candidate pools, effectively narrowing the talent pipeline rather than expanding it. When every company’s AI screens for the same signals, the same people get hired — and the same people get overlooked.
Shapiro’s argument extends beyond strategy into organizational culture. He contends that AI dependency is weakening the muscles of critical thinking within leadership teams. When a senior executive can ask an AI system to generate a market analysis, a strategic recommendation, and a board presentation in minutes, the temptation to skip the hard cognitive work of forming an independent viewpoint becomes overwhelming. The AI doesn’t just assist thinking. It replaces it.
This is where the groupthink risk becomes most acute. Irving Janis, the Yale psychologist who coined the term in 1972, defined groupthink as a pattern of thought characterized by self-deception, forced conformity, and the suppression of dissent. The classic examples — the Bay of Pigs invasion, the Challenger disaster — involved small groups of people in a room, reinforcing each other’s biases. What Shapiro describes is something new: groupthink mediated not by social pressure but by algorithmic consensus. The AI becomes the dominant voice in the room, and because it speaks with statistical confidence, few dare to challenge it.
“Leaders need to be contrarians,” Shapiro wrote in Fortune. “The best business decisions I’ve seen in 40 years came from people who looked at the data and then trusted their gut to go the other way.”
That kind of conviction is getting harder to justify in boardrooms where data-driven decision-making has become synonymous with good governance. Directors and investors increasingly expect leaders to show their analytical receipts. An executive who overrides an AI recommendation based on intuition or experience faces a difficult question: what do you know that the model doesn’t? The honest answer — sometimes nothing concrete, just judgment — doesn’t play well in a quarterly earnings call.
Yet history is littered with examples of leaders whose willingness to defy conventional data-driven wisdom created extraordinary value. Reed Hastings pivoted Netflix from DVD-by-mail to streaming against the advice of virtually every analyst covering the company. Steve Jobs launched the iPhone without any market research suggesting consumers wanted a touchscreen phone without a keyboard. Howard Schultz brought Starbucks to Italy-inspired espresso culture despite focus groups telling him Americans wouldn’t pay $3 for coffee. None of these decisions would have survived an AI recommendation engine trained on existing market data.
The AI tools themselves aren’t to blame. Shapiro is careful to make this point. The technology is extraordinarily capable at pattern recognition, data synthesis, and scenario modeling. Used properly — as one input among many — it can sharpen decision-making considerably. The danger lies in the organizational dynamics that develop around the technology. When AI becomes the default authority, dissent becomes costlier. When every competitor uses the same tools, differentiation requires the courage to ignore them. And courage, as Shapiro notes, is not something you can automate.
Some companies are beginning to recognize the risk. JPMorgan Chase CEO Jamie Dimon has spoken publicly about ensuring that AI augments rather than replaces human judgment in the bank’s operations. In a recent shareholder letter, he emphasized that the bank’s competitive advantage ultimately rests on “the quality of our people’s thinking, not the quality of our algorithms.” Microsoft CEO Satya Nadella has similarly stressed the concept of “AI as copilot” — a framing that deliberately positions the human as the pilot in command.
But these are philosophical statements, and the gap between corporate philosophy and corporate practice is often vast. Middle managers under pressure to hit quarterly targets will take the AI’s recommendation if it provides cover. Risk-averse executives will follow the model’s output because deviating from it creates personal liability. The structural incentives, in most organizations, push toward conformity with AI outputs rather than independence from them.
Shapiro proposes several countermeasures. First, he advocates for what he calls “AI red teams” within organizations — dedicated groups whose job is to challenge AI-generated recommendations and stress-test them against alternative scenarios the models might miss. Second, he argues for preserving space in corporate decision-making for unstructured human debate, the kind of messy, contentious discussion that AI systems can’t replicate and that often surfaces the most valuable insights. Third, he suggests that boards of directors need to explicitly evaluate whether their companies’ AI strategies are producing genuine competitive differentiation or merely replicating industry-wide consensus.
That last point deserves particular attention. Boards have enthusiastically embraced AI governance as a fiduciary responsibility, and rightly so. But most board-level AI discussions focus on implementation speed, cost savings, and risk mitigation. Very few boards are asking the harder question: is our AI making us think more like everyone else?
The timing of Shapiro’s argument is significant. CES 2026, which he oversees, showcased more than 4,500 exhibitors, the vast majority of whom featured AI in their products or services. The show floor was a testament to the technology’s ubiquity. But ubiquity is precisely the condition under which groupthink flourishes. When a capability is rare, it confers advantage. When it’s universal, the advantage shifts to those who use it differently — or who know when not to use it at all.
There’s a parallel here to the early days of the internet. In the late 1990s, companies rushed to build websites and e-commerce platforms, often mimicking each other’s approaches because the technology was new and best practices hadn’t been established. The companies that ultimately won — Amazon, Google, eBay — didn’t just adopt the internet. They thought about it differently than their competitors. They made bets that the data of the day didn’t support. They were, in Shapiro’s framing, contrarians.
The AI era demands the same kind of independent thinking, and it’s precisely this kind of thinking that AI itself tends to discourage. The models are trained on historical data. They optimize for patterns that have worked before. They are, by design, backward-looking even when they’re making forward projections. A leader who relies solely on AI for strategic direction is essentially driving by looking in the rearview mirror.
None of this means companies should slow their AI investments. The technology’s capabilities are real, and the organizations that fail to adopt it will fall behind. But adoption without intellectual independence is a recipe for mediocrity. The companies that will dominate the next decade won’t be the ones with the best AI. They’ll be the ones with leaders who know when to listen to the machine — and when to override it.
Shapiro’s warning, ultimately, is about something larger than technology. It’s about the nature of leadership itself. In an age where algorithms can process more data, identify more patterns, and generate more options than any human mind, the irreducible value of a leader isn’t analytical horsepower. It’s judgment. It’s the willingness to be wrong in a way that no model would recommend. It’s the ability to see what the data doesn’t show.
That’s not a capability you can buy from a vendor or deploy on a cloud platform. And it’s exactly the capability that’s most at risk in the current moment.
WebProNews is an iEntry Publication