MIT warns against replacing Gen Z

Entry-level office work is turning into the first real battleground of the AI labor market. That matters because these jobs are not just cheap labor — they are where people learn how the work actually gets done. The gap is that companies can now automate a lot of the beginner stuff faster than they can redesign how humans get trained. That is why Andrew McAfee, a principal research scientist at MIT Sloan, is warning that cutting junior hiring for AI could be a long-term own goal. ### What exactly did McAfee warn about? McAfee’s argument is not really “AI is bad.” It is narrower and more useful than that. He said companies that replace entry-level workers too quickly risk breaking the path by which novices become experts, managers, and eventually leaders. His the routine stuff. ### Why do routine tasks matter so much? Because beginner work is rarely valuable only for the output. It is valuable for the exposure. A junior analyst doing document cleanup, a first-year associate reviewing contracts, or a young developer fixing small bugs is also learning judgment by doing, watching, and getting corrected in context. ### Isn’t AI supposed to help people learn faster? Yes — but that is different from replacing them. MIT Sloan highlighted new research on 187,000 GitHub developers showing that generative AI shifted work toward more core coding and away from project-management overhead. Junior developers saw the biggest impact. That supports a pretty specific conclusion: the job. If the role disappears, the shortcut becomes a trapdoor. ### Why is Gen Z at the center of this? Because younger workers are the ones trying to get onto the ladder in the first place. McAfee also argued that pulling back on entry-level hiring means turning off the “spigot” of the workers most eager to experiment with AI inside organizations. In other words, companies are becoming more AI-native while sidelining some of their most AI-comfortable employees. ### Is there evidence this is already happening? Basically, yes. Dallas Fed analysis from February 2026 says employment in computer systems design has fallen 5 percent since late 2022, and broader declines in AI-exposed sectors have hit younger workers disproportionately. The pattern is still rising, which suggests experienced workers are being complemented even as junior hiring weakens. ### So is the real issue automation or training design? Training design. The catch is that companies are optimizing for immediate productivity, not for maintaining a future talent pipeline. AI is very good at codified tasks — the textbook part. But businesses still need humans who have built judgment through repeated exposure. If firms strip out the beginner bench a few years later. ### What should companies do instead? The practical answer is not to preserve every old junior task forever. It is to keep entry roles, redesign them, and use AI as a teaching tool instead of a headcount substitute. That means letting juniors do more meaningful work sooner, while still giving them proximity to experienced people. Think less “AI or graduates” and more “AI plus apprentices.” ### What’s the bottom line? McAfee’s warning matters because it reframes entry-level jobs as infrastructure. Cut them, and you do not just save money this quarter — you may erase the system that produces your next generation of skilled workers.