HR and people managers should proceed with caution if they want to use AI to improve efficiency and human capital in the workplace, and should take steps to ensure creativity and critical thinking are preserved, new research from the University of Bath School of Management shows.

“AI is widely promoted as a tool that can support employees by improving efficiency, speeding up problem solving and delivering personalised answers but this should not be taken at face value,” said Professor Dirk Lindebaum, author of the study 'On the Dangers of Large-Language Model Mediated Learning for Human Capital’.

“AI has a part to play in building human capital but it is vital to understand that human knowledge is not uniform. It comes in different kinds, some of which may be more compatible with AI than others,” Professor Lindebaum said.

The research team identified two types of knowledge which appeared partially compatible with AI - encoded knowledge, which encompasses rules, procedures, policies, and datasets; and embedded knowledge - essentially digitalised processes and routines.

“AI may support tasks in these areas by updating documents, policies and workflows, or assisting with compliance and this seems an obvious and easy win for a manager. However, it is not without risk - if employees, for example, no longer engage directly with important processes, familiarity and expertise will fade,” Professor Lindebaum said.

The team identified three forms of knowledge incompatible with AI: embodied knowledge - developed through practical, hands-on experience; encultured knowledge - understanding developed through organisational culture; and embrained knowledge - analytical judgment and problem-solving.

“These three forms of knowledge rely on real-world experience, sensory engagement, socialisation and repeated practice. They cannot be learned through exposure to AI-generated text or synthetic training environments,” Professor Lindebaum said.

“If people begin outsourcing thinking, decision-making or interpretation to AI systems, these critical forms of knowledge wither over time and create a dangerous dependency that could possibly compromise an organisation or a company’s profitability,” he added.

The researchers said HR and people managers need to safeguard against those risks by designing work that ensures continued access to first-hand learning and human interaction, such as shadowing or mentoring at work.

Additionally, they said cultural understanding should be underpinned by onboarding, team-based learning, cross-cultural exchanges and leadership modelling. And HR teams should encourage critical thinking and reflective practices as key skills in their employees to drive human-led decision making.

The research team suggested that critical thinking and creative skills could be safeguarded by created protected spaces within workplaces and educational settings - ‘learning vaults’.

Akin to the Svalbard Seed Vault which safeguards biodiversity, ‘learning vaults’ would be safeguarded from the influence of overly automated learning systems to help ensure employees develop adaptive, experience-based knowledge and maintain the reflexive capacities essential for forming human capital.

“In practise, that would mean a social environment in which employees and students learn how to think for themselves together in terms of know-why (e.g., why did the strategic plan fail?), know-how (e.g., how did a lack of local knowledge contribute to the failure?), and know-what (e.g., what are the consequences of said failure?),” Professor Lindebaum said.

“It would mean ‘learning the basics’ about tasks, processes and routines before cognitive offloading from the beginning undermines the ability of employees and students to provide informed answers about these questions. We think that employers and business schools should explore how such vaults can be integrated into roles and learning environments to protect diverse forms of knowledge that might otherwise be eroded by uncritical AI use,” Professor Lindebaum said.