Future of Labour, Organisation and Work 

with Generative AI (FLOW-GenAI)

I launched the FLOW-GenAI initiative in September 2023.  This is the initial presentation of the initiative with the inaugural workshop.

Generative Artificial Intelligence (GenAI) could be one of the most transformational technologies for workers and, in particular, knowledge workers, as GenAI will transform the way we generate a variety of content. GenAI generates content that cannot be distinguished from human work, blurring the definition of work itself. GenAI is a tech which is both generative and conversational. This means that the outputs of GenAI depend upon our interaction with the technology. GenAI represents a profound transformation in human-machine interaction. Work is highly disrupted by Gen AI, however, at this early stage of adoption, nothing is predetermined about the impact of GenAI on work and workers. The way the technology will be deployed will fundamentally change the way we work, produce and create.

The potential impact on the workforce is large. Walkowiak and MacDonald showed that 37% of the working time of workers is exposed to GenAI[1]. We don’t know yet what it means in terms of productivity gains or labour displacements. But for sure, it will be an important shock on the labour market, impacting a large share of workers, tasks and business processes. As GenAI is a general-purpose technology, like electricity or micro-processors, it will impact all industries, and hence workers from all industries.

New opportunities

In terms of employment, productivity gains from GenAI can drive higher employment levels and create new jobs. For example, in terms of productivity gains, GenAI can speed up idea generation and provide admin support. GenAI allows to free up time to allow us focus on value added tasks. A large part of the current literature in economics demonstrate productivity gains, in terms a rapidity to complete a task without deterioration of the output quality. The immediate translation of adoption of generative AI into productivity gains is something that was not observed for previous technologies. For example, during the nineties, the adoption of computers was associated with what we called the productivity paradox (also named Solow paradox, from the name of the Nobel prize of Economics Robert Solow), which means no immediate productivity gains. In the past, technologies required investments in training and organisational innovations to generate productivity gains. For GenAI, the story looks different. Generative AI represents a reset time for innovation and entrepreneurship, since it substantially changes barriers to entry, by offering new capacities to entrepreneurs. There is a huge potential for technologies to drive better inclusion of workers on the labour market[2]. Once again, barriers to access to jobs are changing with generative AI and this technology can be deployed to support inclusion. For example, a migrant not speaking very well English or a worker with dysgraphia could potentially better perform by using GenAI, without being penalized for their difference. GenAI can play a central role for a more inclusive workforce. Finally, GenAI will lower the cost and accessibility to learning and upskilling, potentially transforming how workers develop their skills. For example, Khan Academy is using generative AI to customize tutoring activities at scale, which is a fantastic news for children and parents who cannot afford to pay a personal tutor. The same may apply within workplaces, access to training is changing with GenAI. In other words, GenAI is an exciting reset time for competition, inclusion and productivity.

New risks

However, GenAI also involves new risks. Automation can generate labour displacements and widen workers inequalities. The key question here is whether GenAI might create a gap between worker who use the technology ethically to improve their performance and workers who don’t use it. Moreover, the deployment of GenAI also introduces new AI risks within the workforce. Workers and GenAI represent two distinct but interdependent sides of the production within workplace, that jointly generate a learning externality that drives productivity gains and can potentially spread error risks at the same time[3]. It means that the economic rational driving the adoption of GAI within workplaces cannot be separated from the potential diffusion of new risks. This is why upskilling strategies to raise awareness on AI risks are central when deploying GenAI as well as implementation of risk mitigation strategies. We mapped the potential diffusion of ten risks within the Australian workforce: competition, industrial relation, intellectual property, accountability, physical safety, misinformation, bias, breach to professional standards, cybersecurity, and privacy risks. We found that economic risks (in terms of competition / industrial relations) are the highest in terms of exposure. But their nature is very different from other risks that require to be mitigated. For example, 9% of the time spent by workers completing their tasks is exposed to a misinformation risk. To sum up, opportunities and risks are intrinsically part of the same transformation, and we have to strategically tackle them together.

It is time to act now for a productive, inclusive and responsible deployment of GenAI. The FLOW-GenAI initiative will provide practical guidelines and recommendations for organisations, to rapidly adapt to GenAI and maximise benefits of this transformation.

[1] Walkowiak, Emmanuelle and MacDonald, Trent, Generative AI and the Workforce: What Are the Risks? (September 12, 2023). http://dx.doi.org/10.2139/ssrn.4568684

[2] Walkowiak, E. (2023). Digitalization and inclusiveness of HRM practices: The example of neurodiversity initiatives. Human Resource Management Journal, 1– 21. https://doi.org/10.1111/1748-8583.12499
Walkowiak, E. (2021). Neurodiversity of the workforce and digital transformation: The case of inclusion of autistic workers at the workplace. Technological Forecasting and Social Change, 168, 120739. https://doi.org/10.1016/j.techfore.2021.120739

[3] Walkowiak, E. (2023). Task-interdependencies between Generative AI and Workers, Economics Letters, Vol. 231, 111315, https://doi.org/10.1016/j.econlet.2023.111315