The conversation around AI often centres on efficiency, with organisations focusing on how quickly they can adopt AI. But the bigger opportunity lies in rethinking how work is designed and where people contribute most effectively alongside AI.
In this episode of the Reshaping Finance podcast, “Rethinking Co-creation at Work: How Human-AI Collaboration Creates Value”, Professor Jasmijn Bol, PwC Professor in Accounting and Francis Martin Chair in Business at Tulane University, explores what building effective human-AI collaboration entails and what you should rethink when AI is part of everyday work.
AI tools can generate content and automate routine tasks, but sustainable business impact comes from determining where human judgement, creativity, and expertise matter most. Organisations must now decide how to allocate work between people and technology to achieve the strongest business outcomes.
Why AI adoption alone isn’t enough: Rethinking work design
“Organisations often treat AI as a productivity tool that they just layer onto existing work. But the deeper change is when AI becomes a contributor, when there is human-AI collaboration, because this changes the division of labour,” according to Bol.
AI adoption is not simply a technology decision. “Technology determines what is possible … but there's also a work design element to this where the collaboration allows for some tasks to be taken over and some tasks to stay with the human.”
As AI handles more structured and routine work, human impact increasingly comes from interpreting information, challenging assumptions, applying context, and exercising sound judgement.
“When you're using AI, it is very easy to give a prompt that focuses on feasibility, that focuses on clarity in writing. So, you can let the AI take over that part and focus all of your effort on novelty, resulting in a bigger overall quality,” Bol said.
Rather than spending time on activities that can be completed more efficiently with AI tools, Bol suggests actively directing effort towards those parts of the workflow where human skills — questioning, innovating, interpreting data — deliver the most impact. The human edge remains strongest in areas where creativity, judgement, and contextual understanding are needed.
Incentives must evolve to reflect where human effort adds most value
Many organisations continue to evaluate outcomes using traditional performance measures. But the research suggests that assumptions about what drives quality may not hold true once AI is part of the workflow.
“When there was no AI, evaluating performance on this broader overall quality worked best … When there was AI, however, this pattern reversed. Now rewarding novelty produced the highest overall quality,” Bol said.
If AI can support feasibility and clarity, organisations may achieve better overall results by rewarding the dimensions where additional human effort makes the greatest difference, whilst maintaining standards across the whole output. The research makes clear that performance systems need to evolve alongside AI as it becomes part of everyday work.
As AI changes where people contribute most, management systems must evolve alongside it.
Sequence matters: Deciding when AI should take the lead
The research also explored whether outcomes differ when AI generates work first versus when people create the initial draft and AI provides feedback.
“When the AI initiated that email, those emails were rated as having higher overall quality, higher creativity, and this is the kicker, higher personal touch,” said Bol.
However, there was a trade-off. Participants reported lower levels of psychological ownership when AI took the lead. This finding suggests there is no universal approach to human-AI working relationships.
Where the immediate priority is a high-quality output, letting AI generate a first draft for people to refine can be effective, even in tasks requiring creativity and social intelligence. Where strengthening employees’ ownership of the work is also important, beginning with a human draft and using AI for feedback may be preferable.
The key is understanding what outcome matters most and structuring the workflow accordingly.
Designing workflows with purpose in mind, not just speed
For leaders in AI-enabled work environments, the challenges will be to design workflows from first principles, not simply to attach AI to what already exists. “It’s really critical that organisations don’t let AI implementation just happen, but that they are very conscious about how to design the organisation around that implementation,” Bol said.
“When managers are looking at a workflow, they have to think of it as not only what I have now, but, ‘If I had a blank slate, how would I design this?’ It’s an opportunity to really question what you’ve been doing and why, and whether those assumptions still make sense.”
In practical terms, this means:
Defining the outcomes that matter most
Allocating tasks and decision rights between people and AI based on capabilities
Ensuring that incentives and feedback systems encourage the right behaviours, such as creativity, judgement, and collaboration
Regularly reviewing and adapting workflows as technology and business needs evolve
AI is changing more than how work gets done. It is reshaping how expertise is applied, how decisions are made, and where professionals create the greatest impact.
“The objective is not to keep humans everywhere,” according to Bol. “It is to keep human contribution where it matters the most.”
Organisations that simply add AI to existing processes may see efficiency gains, but those that rethink how work is organised are more likely to achieve lasting business benefits.
For the full conversation and more insights, listen to the entire conversation on the Reshaping Finance.
Additional resources:
Reimagining Performance: Incentive Design for Human-AI Collaboration
AI and the Changing Way of Working Part 1: How AI Redefines Management Accounting
Real-life Ways Small Firms Use AI
AI in the Public Sector: A Force for Good? Or an Evolving Cautionary Tale?