Organisations are embedding AI into finance functions, but many have not changed the way performance is measured and rewarded.
Incentive systems are powerful levers, shaping behaviour and influencing which areas employees will prioritise. When an organisation is still incentivising outputs and behaviours designed for a pre-AI world, it creates a disconnect between how value is created and rewarded.
The gap between the potential of AI tools and real-world results often comes down to this misalignment: Incentives haven’t changed to recognise where human judgement and creativity now add the most value.
The research report, Incentive Design for Human–AI Collaboration, explores the challenges and solutions to a question many finance leaders face: “How should we measure and reward performance in an AI-enabled world?” The findings are crucial as you work to ensure that your teams are not just using AI but using it to its full potential whilst you oversee performance metrics, KPIs, and reward schemes.
Targeted incentives outperform balanced ones
Traditional management advocates balanced performance measures that reward strategically important dimensions like clarity, feasibility, and originality to ensure well-rounded results. In human-only environments, where employees are responsible for every aspect of their output, this way of measuring performance makes sense.
When AI can handle tasks such as drafting, structuring, and refining work, human effort becomes more valuable elsewhere. The research found that organisations achieve better results when they encourage people to focus on the areas where judgement, creativity, and insight matter most, rather than rewarding every dimension of performance equally.
In controlled experiments, the researchers tracked how professionals interacted with AI. The findings suggest that when incentives focused on novelty and originality, participants used AI differently. Instead of relying on AI to generate ideas, they used it to explore and develop directions they had helped define themselves. Their prompts were more exploratory, and they devoted significantly more attention to expanding the idea space.
By contrast, when incentives focused on overall quality, participants spent more of their interactions on refining outputs, improving structure, strengthening feasibility, and polishing language. In these cases, AI functioned primarily as an editing and optimisation tool.
According to the research, incentives influence how people collaborate with AI. Novelty-focused incentives encouraged participants to use AI in ways that complemented uniquely human strengths, whilst balanced incentives directed more effort toward dimensions that AI already supports well.
Key takeaways and solutions: Aligning incentives with AI-enabled work
The findings from the research raise some important questions and considerations, highlighting the need to rethink how performance is measured, rewarded, and managed.
1. Distinguish between performance floors and differentiators.
Treat AI-supported dimensions (clarity, structure, basic analysis) as baseline expectations — these are the new minimum standards.
Focus human effort (and incentives) on areas where judgement, creativity, and contextual understanding are required.
2. Redesign KPIs and scorecards.
Reweight performance measures to prioritise human-critical contributions, including interpretation, critical thinking, creative problem-solving, and contextual judgement.
In AI-enabled roles, traditional balanced scorecards may need to be revisited.
3. Align rewards with the evolving division of labour.
Bonus schemes and performance reviews should reflect the new reality: AI can support routine tasks, but humans are essential for judgement, originality, contextual understanding, and insight.
4. Think beyond tools and treat AI integration as a management control redesign.
Truly maximising the value of AI requires not just new technology, but rethinking measurement and reward systems.
Putting the research findings into action can look like:
Auditing your current KPIs: Are you still rewarding effort in areas where AI now sets the standard?
Reweighting your scorecard: Shift the focus toward creative problem-solving, professional scepticism, and judgement.
Refining incentive schemes: Consider targeted incentives for innovation, interpretation, and insight—not just output volume or report quality.
As a management accountant, you translate strategy into measurable action. Because AI is reshaping professional work, the systems that manage performance need to change and evolve, too. When you design effective incentives, it redirects effort toward areas where human expertise creates the most value.
Explore the report Incentive Design for Human–AI Collaboration for more details, examples, and actionable frameworks.