AI’s long-term impact on IT jobs still unclear, Microsoft study suggests

There’s mixed news for anyone worried that AI is about to wipe out a swathe of today’s well-paid IT jobs: according to a Microsoft Research study of real-world Copilot use, IT roles will be among the most affected by the technology.

What’s less clear is whether applying AI to IT will mean that specific roles disappear. It’s just as likely that AI will change the way IT people do their jobs rather than remove the need for them, the study suggested.

Being Microsoft’s inhouse chatbot, it makes sense that the authors of Measuring the Occupational Implications of Generative AI would choose to study Copilot’s effects. Since it’s integrated into Microsoft 365, the Edge browser, and the Bing search engine, it is probably the most widely used AI in offices across the US and beyond.

The study analyzed 200,000 anonymized conversations between US users and the Bing Copilot during 2024, in combination with 100,000 interactions drawn from the Copilot user feedback database. From this, the team calculated how Copilot was being applied across a wide range of job roles, relating this to how effectively it performed specific tasks within these roles. 

As one might expect, computing roles scored highly for AI applicability. Put another way, AI can be used to do a lot of things in IT that are done by people today. However, there are important caveats. The study only looked at data for Copilot accessed through Bing search. Other LLMs might have had a much greater or lesser impact.

Similarly, analyzing occupations by breaking them down into individual work activities only gives a partial view of what a job entails. Many roles assume skills and actions not explicitly mentioned in a job description.

The issue of how Copilot is used is also worth underlining. It is a tool used predominantly by office workers, so tells us little about the effect of AI on coding, IT support, or cybersecurity, all of which could be hugely affected by more specialized tools.

What is not clear is whether automating a job role makes it obsolete. The researchers offer the interesting example of the advent of ATMs in the 1970s, which should have reduced the need for bank tellers. In fact, the number of bank tellers increased as banks opened more branches and started using tellers to perform more complex customer banking tasks beyond simple deposit processing.

“It is tempting to conclude that occupations that have high overlap with activities AI performs will be automated and thus experience job or wage loss, and that occupations with activities AI assists with will be augmented and raise wages,” said the researchers. “This would be a mistake, as our data do not include the downstream business impacts of new technology, which are very hard to predict and often counterintuitive.”

So, automation reduced the need for one function, thereby allowing employees to do something completely different. This serves as a reminder that job roles aren’t simply created and retired. Work, and the organizations employing people, constantly evolve to meet pressures other than simple cost calculations.

“This is not a new phenomenon: the majority of employment today is in occupations that arose in the last 100 years as a result of new technologies,” said the researchers. It’s possible that AI will turn out to be a similar story, just as cybersecurity, previously not even a blip in the job market, has come to the fore over recent years. Now the annual ISC2 Cybersecurity Workforce Study routinely reports large skills gaps based on both the unfulfilled demand for people with the right skills and the number of skilled employees organizations should be hiring to meet the underlying need.

It’s been like this across many IT sectors for as long as anyone can remember. Arguably, then, AI won’t reduce demand so much as change the skills needed. The ability to automate job tasks using AI will not end up being the enemy of tech jobs but essential to their future.