Corporate AI Pilots Aren't Scaling, Creating 'Pilot Purgatory'

Companies are running dozens of generative AI pilots but struggling to turn any of them into production systems, according to futurist Amy Webb, who consults with over 100 CEOs annually. She describes the current state as "pilot purgatory"—a cycle of experiments that produce impressive outputs but never integrate into actual business operations.

Webb, who founded the Future Today Strategy Group and teaches at NYU's Stern School of Business, told Fortune that executives are experiencing something analogous to dating-app fatigue. Just as dating platforms profit most when users never find a match, AI pilots seem designed to perpetuate themselves rather than resolve into deployed solutions.

Why it matters

This pattern suggests the corporate AI boom may face a reckoning sooner than Wall Street expects. Companies are increasing AI budgets despite failing to hit cost-saving targets, creating a mismatch between investment enthusiasm and measurable business impact. Webb predicts cracks could appear as early as next year when investors demand results from all those pilot programs.

The hidden costs of AI abundance

Webb's central observation: AI makes production cheap but makes everything else more expensive. One client ran 14 or 15 generative AI pilots since the start of the year using Amazon's two-pizza team rule. None scaled. The problem isn't the technology—it's that pilots run without integration into legal and IT systems, forcing companies to restart from zero each time.

A June survey by Bain & Company of 951 global companies found nearly 40% measured AI cost savings below 10%, despite targeting 11% to 20% returns. Yet 90% are increasing AI budgets anyway, suggesting companies are chasing the sensation of progress rather than actual outcomes.

Decision paralysis from instant analysis

Beyond failed pilots, Webb identifies a second problem: "insta-decks." Presentations that once took a week now take a day, but teams receive five times as many. Executives report direct reports experiencing decision paralysis not from too little information but from drowning in AI-generated analysis they can't process.

"The more a company uses these tools, the more generic ideas are spit out," Webb said. She asks nearly every CEO: if AI freed up 10% of your total capacity tomorrow, where would you deploy it? She hasn't gotten a clear answer yet.

A generational mismatch

Webb sees a structural problem: boards and executives under pressure to adopt AI immediately spent decades developing expertise in unrelated fields. "No CEO was hired because they're an expert in artificial intelligence," she noted. AI is particularly challenging because it's not one technology but an umbrella term for many, requiring data-driven planning rather than executive intuition.

Recent research on "cognitive offloading" shows that when AI handles core reasoning tasks, people's sense of ownership over the work declines. Webb counts on one hand the companies that have figured out sustainable AI experimentation.

Webb expects the reckoning could begin next year as Wall Street starts demanding measurable results from enterprise AI investments. But she resists calling it a simple bubble. "This is not a normal dotcom bubble and burst," she said. "All of this abundance and productivity comes at a new cost that people aren't factoring in."

These details were first reported by Fortune in an interview with Webb.