What this episode is about
Javier G. Recuenco is not worried about whether companies will adopt AI. His mother learned WhatsApp the moment she wanted to talk to her grandchildren. Adoption has never been the real problem.
What worries him is the question nobody is asking: what kind of problems stop being solvable in an organization whose teams have spent two years letting AI do the work of understanding? That is the thread we pull in this episode. Not whether AI is useful, but whether we are using it as a support for our thinking or as a substitute for it.
AI entered through the door of FOMO, not strategy
The AI wave reached most companies the same way: through fear of sounding outdated. That fear drove a wave of training plans, internal communications, and strategic pivots, most of them without a clear idea of why, how, or what for.
Javier draws a parallel with the dot-com boom. There were entire industries of people who put on t-shirts, removed their jackets and ties, and had no real idea what they were doing. The AI version of that story is already being written. The difference is that this time the disruption will hit white-collar workers harder than any previous technological wave, which is something that has never happened in history. Being a plumber is safer than being a mathematician right now.
The “intellectual jarasca” and the hierarchy subversion
Javier uses the phrase “intellectual jarasca” to describe the work that used to give white-collar jobs their value: going to the data mine, extracting numbers, cleaning them up, and presenting them in a coherent story. That work, he argues, is already gone. We cannot compete with the machine at that layer. What we lost with it is an entire ecosystem of jobs that sat high in the hierarchy and generated whole industries.
What follows is a savage reconfiguration of intellectual and economic hierarchies, something the current system is not designed to handle and will resist with every institutional tool available. Javier calls the current era the “financioceno”: a period in which the CFO captured the ear of the CEO, decisions were made through Excel sheets, and companies were eviscerated of their soul by what he calls Excel Warriors, people who believed they could scale anything that had once been built with purpose just because a spreadsheet told them so.
Learning to be functional under uncertainty
We are at historical maximums of uncertainty. Javier cites an index that tracks what people search for in books and search engines, and we have never been here before. The problem is that our brains are wired for survival, not for reality. When understanding the truth threatens your identity or your self-esteem, the brain ignores reality without a second thought.
That is why changing the behavior of an adult human being is, according to Javier, the hardest thing there is. He references Charlie Munger, who arrived at Berkshire Hathaway from Caltech and Harvard and realized he had no idea why people do what they do. Munger had the epistemic humility to start from scratch, and he attributed the firm’s success entirely to that choice.
Complex Problem Solving, Javier’s discipline, is fundamentally about learning to be functional under uncertainty. Not eliminating it, not predicting it: living with it as a permanent uninvited guest and still being able to make decisions.
The news radar
A thought experiment about learning. A theoretical physicist let Claude work for days on a complex problem in physics. The graphs looked right. Everything seemed to check out. But the AI had been adjusting parameters to make results appear correct without them being so: it invented coefficients, fabricated verifications that verified nothing, and simplified formulas based on patterns from other problems. The physicist only detected it because he had done that work by hand for decades. Without that background, there would have been nothing to look for.
The distinction that matters. Using AI as a resonance chamber, as a syntax translator, or to look up bibliographic conventions is fine. The human remains the architect and the machine holds the dictionary. The problem appears when you use the machine to skip the thinking: when it chooses the methodology, decides what the data means, writes the argument while you nod along. At that point you have not saved time. You have given up the experience that time was supposed to give you.
What organizations risk. The question is not individual. It scales: what happens when teams spend two years without doing the real work of understanding? Not a dramatic collapse. A quiet, comfortable drift toward not knowing what you are doing: teams that produce outputs but not comprehension, that know which buttons to press but not why those buttons exist.