Much of the debate about Generative AI focuses on the future. Will it replace jobs? How much productivity will it create? Which industries will be disrupted first?
For many organisations, however, these questions are already being overtaken by reality. Employees are using tools such as ChatGPT, Gemini and Copilot every day.
The more pressing question is: what are people doing with these tools now, and how is AI changing the way they work?
This question was explored through a research project developed from the dissertation of Imelda Adhitama, a student in the University of Edinburgh's MSc Entrepreneurship and Innovation programme (2024-25 cohort). Working with her supervisor, Professor Raluca Bunduchi, Imelda examined how employees use Large Language Models (LLMs) such as ChatGPT and Microsoft Copilot in their daily work, and how different ways of working with AI shape work outcomes. The research was subsequently accepted for presentation at the EURAM 2026 Conference, one of Europe's leading management conferences.
Aligning tasks, AI and outcomes
As GenAI becomes more embedded in organisational work, attention is shifting from what the technology can do to how people can use it effectively.
Drawing on interviews with employees across a large financial services organisation, our research found that outcomes depend on the alignment of three elements: the task being performed, the way users interact with AI, and the prompting strategies they employ. Rather than viewing AI as simply automating or assisting work, our findings reveal a much more nuanced picture.
Employees used GenAI for three main purposes:
- Content generation, including drafting emails, creating presentations, writing code, translating text and developing marketing materials.
- Decision support, such as summarising information, conducting research, evaluating options and exploring alternative solutions.
- Learning and professional development, using AI to explain unfamiliar concepts, simplify technical information and support knowledge acquisition.
But understanding what we do with AI tells only part of the story.
What matters most is how employees work with it.
Our research found little evidence that employees simply hand work over to AI and accept the result. Instead, most uses involved collaboration between human expertise and AI capabilities. Employees refined outputs, challenged suggestions, assigned expert roles to AI, and engaged in iterative conversations to improve outcomes. Even when tasks appeared highly automated, users felt responsible for validating outputs and ensuring quality.
In practice, AI use spans a spectrum. At one end, employees delegate well-defined tasks such as formatting data or drafting routine communications. At the other, they co-create with AI to develop ideas, solve problems and explore alternatives. Between these extremes are multiple forms of human-AI collaboration, from using AI as an alternative source of advice to treating it as a critical reviewer or expert adviser.
The benefits are substantial. Employees reported saving time, improving task quality, generating new ideas and accessing information more effectively. Many also used AI to support learning and broaden perspectives, making it a valuable tool for both productivity and professional development.
However, these benefits came with challenges. Some employees worried about becoming too reliant on AI, while others experienced information overload when presented with too many AI-generated options. There were also concerns that turning to AI rather than colleagues could reduce opportunities for informal learning, mentoring and knowledge sharing.
For organisations, the implication is clear: the greatest value from GenAI may not come from doing existing work faster, but from helping employees develop new ways of working with AI effectively.
This requires investment not only in technology but also in people's capabilities. The most effective users are not simply skilled prompters. They understand what AI can and cannot do, adapt their interactions to different tasks, and critically evaluate whether AI-generated outputs are fit for purpose.
From research to classroom to practice
This project illustrates how Master's dissertation research can contribute to emerging debates that matter to organisations. By connecting academic theory with workplace challenges, students can generate insights that are both academically rigorous and practically relevant.
As organisations continue to experiment with GenAI, understanding how people and AI work together may become a defining capability of the modern workplace. Our research offers an early framework for helping managers think about that challenge.
Reference
Adhitama, I. & Bunduchi, R. 2026. Working with GenAI: Mapping the use of LLM tools in organisational work through aligning tasks, modes of use and outcomes, presented at The EURAM 2026 Conference. Kristiansand, Norway, 16-19 June.