Should promotion depend on how workers use AI?

It's in Duncan Trevithick's interest to be good at using AI at work - if he can use it to get more tasks done, faster, he'll qualify for a bonus at the end of the year.

On the surface, it's a great incentive. But he's questioning what's really in it for him.

"The uncomfortable interpretation is that employees are being assessed on how effectively they can participate in their own redundancy," says Spain-based Trevithick, 34, who works in marketing for an AI training data company.

But if he uses AI to take care of two days worth of work each week, he says the financial rewards don't add up.

"I do not receive two days off or a 40% pay rise. The higher output simply becomes the new baseline. In the short term, that may help me get promoted. In the longer term, I have helped prove how much of my job no longer requires me," he says.

"I'm just trying to view myself as a manager, and I'm managing some people, and I'm also managing AI agents, or AI loops, or whatever you want to call it."

Using employees' AI abilities as a measure of whether they should get a bonus, promoted - or even sacked - is fast becoming more common.

Accenture CEO Julie Sweet told the Rapid Response podcast in March: "Today, AI at Accenture is how we do work. So if you want to get promoted, you've got to do the things that we do in order to operate at Accenture."

And big names like Disney, Meta, JP Morgan and KPMG have introduced "AI leaderboards" to track and rank employees' usage of the array of LLMs and platforms at their disposal, according to media reports. Crypto trading platform Coinbase has already fired engineers who didn't complete AI training, as requested by the chief executive, Brian Armstrong.

It demonstrates how eager business leaders are to see a return from their AI investments, especially when 94% of companies have yet to see significant value from AI, according to a report by consultants McKinsey.

Employees feel they have to get on board, thanks to the current state of the jobs market - in the UK, vacancies have hit a five-year low.

Meanwhile, 75% of the 1,881 UK jobseekers polled in July by recruiter Gi Group say they would not be put off applying to an organisation that had introduced AI proficiency into individual performance reviews. Around 22% said it would be off-putting.

"It's like being in the sea and you see a giant wave coming. You can get out a surfboard and try and ride it as long as possible, or you can just let it take you under. But you can't stop the wave," says Trevithick.

But what's going on at the large consultancy Pamela (her name has been changed to protect her identity) works for isn't sitting right with her.

She's a US-based senior executive who has observed that, while there isn't a formal mandate for employees to use AI, it's obviously determining who is rewarded, and who is left behind.

"The ground is shifting under us. You've got to demonstrate [AI] fluency and fluidity as one of your key achievements. It's kind of quiet where nobody's saying, 'learn AI or else'. But let me tell you, in performance reviews, they reward who uses it well," says Pamela.

That shift, she adds, means that AI capabilities are now counting for more than actual experience, which is creating a "two tier workforce".

"You've got the same job title, same tenure, but different value based on whether someone treats AI as a threat or a tool - that gap widens really quickly once leadership notices it," says Pamela.

"AI fluency beats credentials every day. Somebody that has 15, 20 years of experience and no AI fluency will be passed over for those that have, say, three years, but are fast with the tools."

She adds: "If you're not visibly using AI, you see slower promotion timelines. It's harder to be seen, and it's harder to fight being on that shortlist."

But is it legal for companies to shift the goal posts like this?

Legally, bosses are in the clear, confirms Tina Chander, partner and employment lawyer at Weightmans. But she says it does open up questions about pay and fairness.

"The question then becomes whether the employer increases expectations on the basis that the employee can now produce more work. Is that fair?" questions Chander.

"And if an employee is effectively doing more because AI has made them more efficient, should they be rewarded differently? Or are they effectively making themselves redundant, thus acting as a disincentive to be productive?"

Chander advises businesses to have clear policies, training and boundaries in place, otherwise they risk employees opening up disputes about fairness, performance expectations and job security.

Indeed, from January 2027, UK employees will have six months to submit unfair dismissal claims compared with the current three months, and will be able to do so after just six months service, rather than the current two years.

Yet HR consultant Tina Rahman says employers aren't currently being transparent enough in telling employees why they want AI to be part of people's jobs, and what the end purpose of that is - which she agrees is to save costs and time, but also to cut down on outsourcing.

"Because they misunderstand it, this is not being reflected to employees," says Rahman, who runs a London-based consultancy, HR Habitat.

A lack of clarity around AI usage is certainly causing discontent at Pamela's company.

"Leadership is being inconsistent and vague on purpose. Firms want productivity gains without owning the disruption narrative," she says.

"If something goes wrong, they're going to say, 'I didn't tell you to do that. Where'd you get that from?' Then if you did something with AI, and it works, great."

Mandating AI use at work can become purely performative, rather than having any real positive impact, warns Kamila Miller, applied AI researcher and lecturer at Henley Business School.

"Make AI usage a KPI [key performance indicator], and people will log their interactions to hit the metric, route work through a chatbot that did not need it, and generate AI-flavoured outputs that look productive on a dashboard. You will measure adoption. You will not measure judgement, learning, or better decisions," states Miller, who is based in Reading.

"You have not made people more skilled - you have made them more obedient."

Some firms have acknowledged those arguments.

In April, Luis von Ahn, the CEO of Duolingo, told the podcast Silicon Valley Girl that his language learning firm is no longer using AI use as part of performance reviews.

"We found that people... were asking: 'Do you want us to use AI for AI's sake?'

"In the end we backtracked and we said: 'Look, the most important thing for your performance is that you are doing, whatever your job is, as well as possible. A lot of times AI can help you with that. But if it can't, I'm not going to force you to do that.'"

Meanwhile, in May, the Financial Times reported that Amazon had shut down an internal leaderboard which tracked employee's use of AI. The report said employees had been attempting to climb the rankings by setting AI needless tasks.

Despite those high-profile moves, some staff still they need to show AI proficiency - Pamela is one of those.

In her mid 50s, she plans to retire within 10 years, and wants to keep her job so she doesn't risk losing any pension and health benefits. She's visibly stepping up how AI is helping her bring in new clients, and making sure she's being credited for it.

As for Duncan Trevithick, he's doing his best to future proof himself with a number of side projects.

"If AI can do a job better than you can, it makes sense for the business to replace you. That's how the capitalist model works. So it's about, how can you move into a position where you own assets where you can leverage AI, and then you benefit," says Trevithick.