Four risks of relying on ‘carrots’ to make AI stick

As tax season arrives, and most companies weigh up their hiring and promotion decisions on the basis of the closing financial year, many staff now have a new metric to worry about: how often they have deployed AI in their work. But even as many employers make statements that those not seen as ‘AI-ready’ will find limited opportunities, Steve Salvin, CEO of Aiimi, cautions that pressuring staff to performatively adopt AI will have lasting consequences on a business’ performance.

It’s April 2027 and time for your year-end review. Your mind is racing to count how often you’ve opened the company’s AI assistant over the last twelve months. How well those numbers stack up against your colleagues’ could determine whether it’s time for a promotion, or a departure.

While it might seem far-fetched, recent overtures from consultancy firms looking to drive up internal AI engagement mean these scenarios could become reality faster than we think.  

It’s only been three years since the launch of ChatGPT ignited a paradigm shift the world over. In that time, consultancies have been at the vanguard of embracing and integrating AI tools - radically reshaping how they work internally and the services they sell externally. And in recent weeks, we’ve begun to see a new chapter unfolding: the push to unlock a new era of company-wide engagement with AI tools. 

Accenture recently made public statements about how they want to drive up staff usage of AI, going as far as saying it will be linked to bonuses and promotions. It’s hoped that offering this type of ‘carrot’ incentive to boost employee adoption of the latest tech tools will improve overall AI usage metrics and harmonise the workforce around a new way of doing things. 

And while there’s logic to offering incentives, such a strategy could also create risks if not rolled out thoughtfully. Over-reliance on carrot (or stick) methods to push teams into the AI age must take into account these four considerations. 

Remove the risk of ‘performative’ AI

Perhaps the most obvious issue to avoid is measuring engagement with one-dimensional data. This could be how often AI tools are opened, how many minutes are spent on various platforms, or the quantity of conversations had with internal chatbots. 

If promotions or bonuses are tied to such metrics, employees won’t be taking the time to really engage with the technology’s possibilities; they’ll be focused on ‘performative’ use of AI that looks good on paper, but doesn’t move the dial in reality. Simplistic quantitative approaches could promote surface-level engagement or create confusion around which actions drive true value for the organisation. 

History has taught us this lesson repeatedly. Ahead of the 2003 World Cup, the England rugby team introduced tech to track matches, including how many meters players ran each game -  creating a live leaderboard with the results. Players like Will Greenwood, whose position meant they naturally covered less ground, took to doing shuttle runs during stoppage time - boosting their ranking and highlighting the fundamental flaws with a system that hadn’t accounted for nuance.

Dig deeper on the data

Removing surface-level metrics as a benchmark for success is only half the battle. You then need to replace these with processes that encourage meaningful engagement. That means enabling colleagues to really interrogate and explore AI. Give them sandboxes to experiment in, early access to tools you’re considering adopting, and the opportunity to reject what doesn’t seem to be working. 

KPMG is pioneering some compelling examples in this regard: rather than tying promotions to surface-level metrics, the firm is offering cash prizes for employees who show true innovation with AI. PwC has also echoed this ‘dig deeper’ narrative, saying staff bonuses would be based on revenue and margin metrics, rather than volume of AI usage. Going deeper and building more nuance around the methods of tracking engagement is a far better way to ensure ROI. 

Don’t underestimate the value of ‘refuseniks’ 

Times change and employees have to change with them. But unwittingly pushing all staff to engage with AI tools in a uniform way could not only generate resentment, but also mean enterprises lose out on valuable opportunities to iterate their strategies. 

There are many reasons staff might not be using your AI tools – and these may have nothing to do with general tech-reticence. It could be that your training hasn’t been thorough enough. Perhaps specific systems work really well for one area of the business, but aren’t fit for purpose in another. Or maybe there’s a lack of the good quality, well-governed data needed for AI success. The lure of a bonus could discourage those running into friction from speaking up about these issues. 

Rather than introducing incentives that could unwittingly mask potential AI pain points, enterprises should incentivise feedback. Speak to your teams - including those who seem the least enthused by your existing AI efforts - and gather opinions from the frontline. Understand what’s working and what’s creating frustration. Use this to refine efforts going forward, before you rush to reward those for whom AI fits most seamlessly into their day-to-day. 

AI is most effective when moulded to your employees, not when they are moulded to it.

Beware of the ‘stick’

While a well-designed carrot can unlock impact, the stick approach is rarely popular with staff. When it comes to AI roll-outs, we must prioritise instilling hope and curiosity rather than fear. If we don’t, the quest for operational efficiencies risks alienating the talented people at the heart of our enterprises.

Of course, sometimes teams need a push to embrace new ways of working. But it’s vital that they feel brought into the logic and the process: blunt measures will rarely lead to sharp returns.

In my decades spent working with AI, one constant has remained: innovation cannot truly embed without the goodwill and energy of dedicated staff. Before enterprises can reap the very real rewards of AI roll-outs, they need to ensure they have a robust, fair, and inclusive roadmap for bringing colleagues on the journey with them.