• AI is going to have a huge impact on the workforce, making many workers more productive.
  • But just as there will be winners in the workplace, there will be losers.
  • It looks like AI will negatively impact mid-career, mid-ability white-collar workers the most. 

"Mediocrity will be automated."

That was the verdict a top tech executive shared with me recently, describing the impact he predicted AI would have on the workforce. And while the phrasing might seem a bit harsh, there's growing evidence that he might be on to something.

More specifically, AI could disproportionately impact the middle class of white-collar workers — the folks who are mid-career, mid-ability, mid-level, and yes, in some cases, mediocre. Here's why:

AI benefits the least experienced workers

Academics Erik Brynjolfsson, Lindsey R. Raymond, and Danielle Li recently studied the impact of access to an AI-based conversational assistant on almost 5,200 customer support agents at a Fortune 500 software company. The trio found that the tool helped increase productivity by 14%. And critically, it was novice workers who benefited most.

"In contrast to studies of prior waves of computerization, we find that these gains accrue disproportionately to less-experienced and lower-skill workers," read their academic paper. "We argue that this occurs because ML systems work by capturing and disseminating the patterns of behavior that characterize the most productive agents."

In other words, the lessons learned from months or years of experience are baked into an AI tool. As novice workers get access to these tools, they're supercharged, helping them close the gap in performance with more experienced colleagues. 

Or, as the study puts it: "We find that the AI tool helps newer agents move more quickly down the experience curve: treated agents with two months of tenure perform just as well as untreated agents with over six months of tenure."

It isn't just customer support work where this dynamic could take hold. Microsoft CEO Satya Nadella, for example, recently told Time that the same could be true for software developers. He said: 

I mean, to give you a concrete example, developers who are using GitHub Copilot are 50-odd percent more productive, staying more in the flow. We have around 100 million professional developers, we think the world probably can get to a billion professional developers. That will be a massive increase in total developers, because the barriers to being a software developer are going to come down. This doesn't mean the great software developers won't remain great software developers but the ability for more people to enter the field will increase.

That's good news for many wannabe software developers, but it's also bad news for many existing software developers. That's because lowering the barrier to entry will increase the supply of workers, putting downward pressure on wages. 

(For more on how AI could impact software developers, I highly recommend this story from my colleague Aki Ito on "the end of coding as we know it.")

The middle class of knowledge workers will face their own Uber moment

Professor Carl Benedikt Frey, the director of future of work at the Oxford Martin School, likens this shift to the impact of Uber on London taxi drivers. For decades, every taxi driver in London has been required to pass a test called the Knowledge, which requires memorizing miles of streets of central London.

Then, Uber arrived and essentially put the Knowledge on every cell phone in every car in London.   

"Suddenly, knowing the name of each street in London was no longer valuable expertise, so that anybody with a drivers license could drive a taxi," Frey told me over email. "The result was more competition for incumbent taxi drivers who saw their incomes fall by around 10%."

Now you could see the same effect on lots of white-collar work. Think translators, web designers, lawyers, coders, accountants, copywriters, or HR professionals. The skills developed through advanced degrees or years of experience in a specific role or company might soon be embedded into a generative AI tool, lowering the bar to entry. 

The benefit to employers here is clear. If a less experienced employee aided by an AI tool can be as effective as a more expensive, more experienced employee, you can guess what's going to happen next. 

A version of this has already played out in many places. Take Wall Street trading floors, for example. Whereas top traders and salespeople were employed in the past because of their market savvy and experience, banks have, over the past decade, found ways to lower their costs by moving out many of those types, and replacing them with junior employees with better tech tools. There was even a word for this process of replacement: juniorization. 

Now you could see a similar, accelerated process play out across many more white-collar industries.

That's before you get to the potential for workers to be replaced entirely by AI. I wrote back in April that tech companies weren't just cutting jobs, but that many of those jobs weren't going to come back. Within weeks, IBM's CEO said it was halting hiring for almost 8,000 jobs that could be replaced by AI. British telecommunications giant BT Group latest said it was cutting 55,000 jobs, and that AI could replace 10,000 of those by 2030. 

If you're worried about what AI means for your career, start using it

To be sure, there's lots that's still unclear about how all this plays out. AI tools are still in their diapers, and we don't yet know what they'll grow up to be. And much will depend on how companies decide to employ AI.

"Workers' experiences may be more dependent on the nature of their firms' adoption of these technologies," Mark Muro, a senior fellow at the Brookings Institution, told my colleague Jacob Zinkula. "Are they adopted in a spirit of improving processes and creating more and better output? Or are they used crudely just to get rid of workers? That's a difference between 'good AI' and 'bad AI.'"

In a Microsoft study back in February and March, 31% of business leaders said they saw increasing employee productivity as what they would value most about AI in the workplace. 29% mentioned helping employees with necessary but repetitive/mundane tasks, and 25% said eliminating employee time-spend on low-value activities. 

16% said reducing headcount. 

If you're worried about what AI means for your job, the best defense is to learn how to use AI to your advantage. 

"AI won't take your job," Richard Baldwin, an economist and professor at the Geneva Graduate Institute in Switzerland, said at a recent event, per my colleague Aaron Mok. "It's somebody using AI that will take your job."