The claim that computer science is a dying degree keeps circulating, fed by stories of graduates working at Chipotle and headlines about a bursting bubble. The real picture is messier than either the doom story or the reassurance piece written to counter it.
Start with the honest bad news. Government wage data really does show computer science graduates facing a higher jobless rate than most majors, somewhere around 6 percent, worse than art history and dozens of other fields. That part of the pessimistic story is true, not exaggerated.
The reassurance comes from a different number. When computer science graduates do have jobs, they are less likely than most other majors to be stuck in work that does not need a degree at all. And on pay, computer science still tops the charts. Projected starting salaries for this year's graduating class remain the highest of any major, averaging in the eighties of thousands of dollars for bachelor's degree holders.
But averages hide the real fracture line, which is age and seniority, not the degree itself. A Stanford study found that employment for software developers in their early twenties has fallen by almost 20 percent since ChatGPT was released to the public. That is a specific, measurable drop, and it matches what recruiters have said for two years: junior coding work, the kind fresh graduates used to be hired to do, is exactly the kind of task AI tools now handle well.
What is growing instead is a different job entirely. Companies are increasingly hiring people to sit inside a business, learn how that business actually runs, and then use AI tools to rebuild the workflow around it. This role, often called a forward deployed engineer, started at the data company Palantir and has since spread to consulting firms and other AI companies, with hiring demand for it climbing sharply over the past year.
This matters well beyond the tech industry. The same pattern playing out in software hiring, cheap automated tools eating routine work while demand rises for people who can direct those tools inside one specific business, is coming for other technical and semi-technical roles in insurance, logistics, and manufacturing too.
The practical takeaway is not to stop hiring or encouraging computer science graduates. It is to change what gets evaluated. A degree alone no longer tells an employer much on its own. What matters more is whether a candidate has done real hands-on work, an internship, a project, a certification, that shows they can apply AI tools to an actual business problem rather than just knowing how to use one.
Universities are already adjusting. CUNY has pushed new AI coursework across its campuses, and Howard University is rolling out an AI certificate program this year, both signs that schools see where hiring is actually heading, even if the public conversation has not caught up yet.