Where is AI genuinely creating value, and where is it just noise? To open Unscripted with Jeff Pedowitz, I put that question to someone who has spent a career building things that have to work on the first try, in places where a second try is not an option.
The short answer from that conversation: AI is real, but most organizations are not getting value from it, and the reason is rarely the technology. It is that they treat AI like fairy dust sprinkled on a workflow they never actually changed. AI creates value when you take a process apart, find the friction, and apply the right method to that specific point. It does not create value as a label you attach to what you were already doing.
Zaheer Ali is a research physicist turned AI entrepreneur. He has built instruments for NASA, run shots at a laser fusion lab, and helped found a national center for nuclear security. Today he is building the first true Space MBA, helping shape an industry that could approach $2 trillion within a decade. He works at an intersection almost no one else occupies, connecting artificial intelligence, quantum, and space, and he approaches all of it as a practicing Stoic who draws on Eastern, Western, and Indigenous ideas in roughly equal measure.
That background is the reason his read on AI is worth listening to. He is not selling a model. He is a scientist who has shipped hardware into environments that punish overconfidence.
One of the first things Zaheer pushed back on is the idea that we are witnessing the arrival of something that did not exist before. The mathematics behind much of today's AI has been around for decades. What changed is the maturity of the methods and the scale of compute now available to run them.
This reframing is useful because it moves the question from "is the technology ready" to "are we using it correctly." When AI fails to deliver inside a company, the failure usually sits in how it was applied, not in the model itself.
Zaheer's memorable phrase is that most businesses treat AI like fairy dust. They sprinkle it on an unchanged process and expect a different outcome. That is not how transformation works. Every major wave of change, from the internet to mobile to cloud, was a culture shift first and a technology shift second. The companies that win with AI are the ones willing to change how work actually gets done, not just to add a tool on top of it.
The practical takeaway for leaders: if your AI initiative leaves your workflows, incentives, and decision rights exactly as they were, you have bought fairy dust.
Zaheer grounded the conversation in two concrete domains.
In materials discovery, AI compresses a search space that used to require years of slow, expensive physical experimentation. Instead of testing candidates one by one, you narrow the field computationally and run experiments only on the most promising options.
In mergers and acquisitions, AI changes the economics of evaluating a target: how you assess value, where risk is hiding, and how quickly you can get to a defensible decision.
The common thread in both is discipline. You take the value chain apart, locate the friction, and apply the right kind of AI to that specific point. Not the flashiest model. The one that fits the problem.
Zaheer is skeptical of the certainty in the artificial general intelligence and singularity narrative. He is not dismissive, he is skeptical, and the distinction matters. As a scientist, he wants evidence proportional to the claim, and he does not believe the current evidence supports the confidence with which some people forecast the arrival of general intelligence.
For business leaders drowning in bold predictions, that humility is worth borrowing. You can invest aggressively in what AI does well today without betting your strategy on a specific date for machine superintelligence.
This is where the Stoic thread runs underneath the entire episode. Zaheer's framework for governing powerful technology is not about eliminating risk, which is impossible. It is about minimizing harm and refusing the two easy extremes of hype and fear.
When you build for deep space, you learn to plan carefully for failure without being paralyzed by it. That posture translates directly to deploying AI inside an organization: name what can go wrong, reduce it as far as is reasonable, and act, rather than freezing or pretending the risk is zero.
The frontier Zaheer is watching is AI reshaping the physical world through generative design and advanced manufacturing. Not more screens and chatbots, but the objects, materials, and systems around us being designed and produced differently. Paired with the growth of the space economy, that is a very different picture of AI's impact than the one dominating most feeds.
Is AI overhyped? Both things are true at once. The capability is real and improving, but the value most companies extract is low because they apply AI to unchanged processes rather than redesigning the work. The hype is in the expectation that AI produces results without any change in how you operate.
Why do AI projects fail in large companies? The most common reason is treating AI as a feature added on top of existing workflows instead of as a shift in how work is done. Transformation is a culture change first and a technology change second.
What is the "fairy dust" problem in AI adoption? It is the belief that sprinkling AI onto a process you never redesigned will produce a meaningfully better outcome. It rarely does. Real gains come from taking the value chain apart and applying AI to specific points of friction.
How big is the space economy expected to get? Zaheer works from an outlook in which the space economy could approach $2 trillion within a decade, which is part of why he is building an education pathway focused specifically on that market.
What is a Stoic approach to AI risk? Focus on what you can control, minimize potential harm, and avoid the paralysis of fear and the recklessness of hype. Plan for failure without being frozen by it.
Listen to the full conversation with Zaheer Ali, and catch every episode, on the Unscripted with Jeff Pedowitz page.