Everyone has an AI jobs take. Executives say it'll free us from drudgery. Workers say it'll destroy careers. This MIT study actually looked at what's happening in real workplaces.The data is messy and contradictory, which probably means it's closer to the truth than either camp wants to admit.A new MIT study analyzing actual labor market impacts suggests the AI job apocalypse narrative may be overblown. Researchers found that automation effects are more nuanced than headline-grabbing predictions suggest, with significant job displacement in some sectors but also unexpected job creation in others.Let's be clear about what this study isn't. It's not saying AI won't affect jobs. It's saying the effects are complicated and vary wildly by industry, role, and how companies actually deploy the technology.Some jobs are getting automated. Customer service roles, data entry, basic content writing, they're shrinking as AI tools handle routine tasks. That displacement is real and affecting real people.But other jobs are expanding. AI creates demand for trainers, prompt engineers, quality reviewers, and specialists who can integrate AI tools into workflows. Some displaced workers are moving into those roles. Some aren't.The technology is impressive. The question is whether it creates more opportunities than it destroys.According to MIT's data, it depends. In sectors that adopted AI quickly, like finance and tech, employment actually grew despite automation. Companies used AI to handle increased workload rather than reduce headcount.In sectors with thin margins and intense competition, like call centers and basic content production, AI enabled companies to cut labor costs. Jobs disappeared.The difference is what companies optimize for. Growth-oriented businesses use AI to do more with the same staff. Cost-cutting businesses use AI to do the same with less staff.Both are rational responses. The first helps workers. The second doesn't.MIT's study also found that AI's impact isn't uniform within job categories. Senior customer service representatives who can handle complex issues are in higher demand as AI takes routine calls. Entry-level workers who were learning the role through handling routine calls are losing opportunities.That's the subtle, pernicious effect nobody talks about. AI doesn't just eliminate current jobs. It eliminates the entry path to careers. If junior roles get automated, how do people develop expertise to reach senior roles?The study doesn't have answers. It has better questions.Like: Are companies investing in retraining displaced workers, or just hiring different workers? Are AI productivity gains benefiting employees through higher wages, or just shareholders through higher profits? Is AI creating career ladders or breaking them?The answers vary by company and industry. That's frustrating for people who want simple predictions, but it's honest.Doom-and-gloom predictions of 40% job losses are probably wrong. They assume AI capabilities that don't exist yet and ignore how slowly organizations actually change. Most companies can barely implement email correctly. They're not restructuring around AI overnight.But rosy predictions that AI will and create universal prosperity are also wrong. Displacement is happening now. People are losing jobs. The new opportunities require different skills, and not everyone can retrain.MIT's conclusion is basically: it's complicated, watch the data, and stop trusting anyone who claims to know exactly what will happen.That's not satisfying. But it's more useful than false certainty.The labor market impacts of AI will depend on policy choices, how companies deploy the technology, whether workers have power to negotiate over implementation, and what training resources are available.Those are all variables humans control. AI isn't a force of nature. It's a tool deployed by people making choices.The job apocalypse isn't inevitable. Neither is a worker's paradise. The outcome depends on decisions being made right now by companies, governments, and workers.MIT's study suggests we should spend less time predicting the future and more time shaping it.