The playbook for building judgment at work is about to be rewritten — and it won't involve watching a senior colleague handle the messy client call.
The Summary
- Mark Cuban predicts AI-powered job simulators will replace traditional employee training, comparing future workplace training to how pilots and race car drivers learn through simulated scenarios
- The shift addresses a coming gap: as AI handles more routine tasks, workers will have fewer "touch points" with colleagues to build the judgment that historically came from experience
- Cuban suggests companies will task domain experts with building custom simulators that walk employees through every possible workplace scenario
The Signal
Cuban's prediction lands at a moment when the apprenticeship model of work is already showing cracks. Junior employees at consulting firms, law offices, and creative agencies are finding fewer opportunities to learn by doing because AI agents are increasingly handling the repetitive tasks that once built foundational skills.
The billionaire investor argues that simulation-based training will fill this void. Think of a newly qualified lawyer running through hundreds of deposition scenarios, or a customer service manager practicing high-stakes client conversations with AI-generated personas that adapt in real time. The comparison to pilot training is apt: flight simulators allow for practice of rare, high-risk scenarios that would be dangerous or impossible to rehearse in real aircraft.
"Employees won't have as many touch points in the company to gain knowledge and experience from. That's where judgment has historically come from."
What Cuban doesn't fully address is the economic implication. Traditional training was expensive but distributed across the workforce through mentorship and observation. Simulation-based training shifts costs upfront but could dramatically compress learning curves. A role that once required three years of grinding through routine work to build judgment might require three months of intensive simulation plus immediate deep work.
The open source angle matters here. Cuban specifically mentions "open source, or open weights" as the driver of these simulators. That suggests training environments won't be locked behind enterprise paywalls. Companies could fork and customize base models, building simulations specific to their operations, their customers, their actual problems.
Key shifts this implies:
- Domain expertise becomes more valuable as simulator design, not just as knowledge to hoard
- Entry-level roles shrink further while demand for judgment-heavy positions increases
- Geographic arbitrage in labor markets weakens when anyone can access identical training infrastructure
The risk is assuming simulation fully replicates the messy reality of workplace politics, unspoken hierarchies, and the emotional labor of reading a room. Cuban himself noted earlier this week that AI can't match human empathy or situational awareness. Simulators can teach you how to run a client meeting. They can't teach you what to do when your boss undermines you in that meeting.
The Implication
If Cuban's right, the next five years will split workers into two groups: those who use AI simulators to compress decades of learning into months, and those who treat traditional career ladders like they still exist. Companies already building these training environments have an edge in talent development that competitors won't see coming until it's too late.
For individual workers, the shift means rethinking how you signal competence. Credentials and years of experience matter less when anyone can run ten thousand simulated scenarios in their spare time. What matters is demonstrable judgment under conditions the simulator couldn't predict.