Mark Cuban has never been shy about predicting where the economy is headed. In March 2025, he turned his attention to the office. Sitting down with Shira Lazar at SXSW for Inc., the billionaire investor said the workers who should worry most about artificial intelligence aren't the ones just starting out.
It's the people in the middle, running teams and shuffling reports, who he thinks need to pay attention. That warning lands at a moment when a lot of readers are trying to get ahead financially while AI quietly reshapes what their jobs look like day to day.
Federal labor projections and years of automation research back up part of what Cuban said. They also show the real picture is more layered than a simple countdown to layoffs.
What Cuban actually told Inc.
Picture the manager whose whole day is spent compiling status updates, routing requests, and keeping everyone's calendar straight. That's the job Cuban had in mind when he spoke with Lazar in March 2025.
"It's not the entry-level jobs that are going first, it's middle management," he said, according to the Inc. interview. "If your job is just managing people or organizing data, you better learn AI, because AI is learning your job."
Cuban also framed AI as something workers can use rather than just fear. In the same conversation, he said, "It's not about replacing people, it's about making them more efficient," pointing to AI as an efficiency tool businesses are already putting to work.
Why organizing information looks exposed
Cuban zeroed in on a specific kind of work: organizing data and coordinating people. That lines up with how automation researchers have measured AI's reach for years. McKinsey Global Institute's 2017 analysis found that collecting and processing data were among the most automatable activities in the economy.
McKinsey put the technical automation potential of those tasks at 64% and 69%. By 2023, the firm updated that estimate for generative AI specifically, finding the technical automation potential for knowledge work had climbed from roughly 50% to somewhere between 60% and 70%, with data management flagged for further gains.
Federal projections tell a similar story at the occupation level. The Bureau of Labor Statistics expects employment of secretaries and administrative assistants, a role built on scheduling and routine reporting, to decline 2% from 2025 to 2035.
Management analysts, whose job also involves organizing information but adds interpretation and recommendations, are projected to grow 10% over the same stretch, according to BLS.
What BLS's own numbers show about management jobs
Cuban's warning was aimed at middle management broadly, so it's worth checking that group against the government's own projections. The office and administrative support occupational group handles much of the routine coordination work Cuban described.
That group is projected to shrink 3.9% from 2024 to 2034, a loss of about 761,900 jobs, according to the Monthly Labor Review. Newer 2025-2035 BLS projections put the decline even steeper, at 4.0% and 752,100 jobs lost, the fastest drop of any major occupational group.
But management occupations as a whole are not shrinking. BLS projects the broader management group to grow 6.1% and add 833,400 jobs from 2024 to 2034. General and operations managers are projected to grow 5.0% from 2025 to 2035, and administrative services and facilities managers are projected to grow 5% over the same period.
BLS has been explicit that this is about tasks, not whole jobs disappearing. The agency's Economics Daily report says that as AI integration expands, demand is expected to be limited for roles like procurement clerks, credit authorizers, and legal secretaries, because those workers will see added efficiency gains rather than being replaced wholesale.
Tasks disappear before jobs do
It's easy to hear "AI is learning your job" and picture an entire role vanishing overnight. Researchers who study automation generally don't see it that way.
A widely cited OECD study re-estimated automation risk at the task level instead of the occupation level and found that only about 9% of jobs across OECD countries face high automation risk, far below earlier occupation-based estimates that put the number closer to 47%.
A follow-up OECD analysis using adult skills survey data found 14% of jobs at high risk of automation, with another 32% facing a 50% to 70% probability.
An October 2024 Brookings analysis of OpenAI task-exposure data found that more than 30% of all workers could see at least half their tasks affected by generative AI, and 85% could see at least 10% of their tasks affected.
Brookings was careful to separate exposure from elimination, noting that whether a job actually disappears depends on employer choices and market response, not just what AI is technically capable of doing.
The skills AI still can't touch
If routine reporting is the part of a job most exposed, judgment and relationships are the parts holding up best. The World Economic Forum's Future of Jobs Report, released in January 2025, had GPT-4o assess more than 2,800 granular workplace skills.
None rated "very high capacity" for AI substitution, and 69% rated low or very low capacity, largely because AI still struggles with nuanced judgment and hands-on application. The same report found analytical thinking remains the single most in-demand core skill, with seven in 10 employers calling it essential.
It also found that skills rooted in human interaction, like empathy and active listening, currently show no substitution potential at all. Relationship-building shows up in hiring data too.
LinkedIn named it the top skill on its Skills on the Rise 2025 list, and a 2026 GMAC survey of more than 600 global employers found communication was the single most valuable attribute they look for, far outranking AI tool proficiency.
Bottom line
Cuban's warning is less about a specific job title disappearing and more about which parts of a job are safe to automate. The evidence backs him up on that distinction: routine reporting, scheduling, and data organizing are shrinking in demand, while work that requires analysis, decisions, and relationships is holding steady or growing.
If your job leans heavily on compiling and forwarding information, it's worth learning the AI tools that can do that faster than you can, and shifting your time toward the judgment calls and stakeholder relationships a tool can't replicate.
It's also smart to build some financial breathing room while you make that shift. Strengthening your savings or looking into extra income outside of work can act as a hedge while your role evolves.
The workers least exposed to this wave of automation won't be the ones who ignore AI. They will be the ones who let it take the busywork off their plate so they can focus on the parts of the job only a person can do.
Add Us On Google