With AI systems being more and more used to complement human tasks, the question arises about the changing nature of work: what does AI change to jobs?
The research
Three recent studies have explored the question in different settings: coders, R&D teams and professional writers.
A first study leveraged millions of work activities over a two-year period to investigate the impact of Github Copilot on the task allocation of software developers. The results show that having access to Copilot induces to allocate more time to core work of coding activities (+12%) and less from non-core project management activities (-25%).
A second study analysed the impact of introducing AI to the 1,018 scientists in the R&D lab of a large U.S. firm. In this setting, AI automates 57% of “idea-generation” tasks, reallocating researchers to the new task of evaluating model-produced candidate materials.
In a third study, researchers conducted an experiment with 444 professional writers and observed that Generative AI restructures tasks towards idea-generation and editing and away from rough-drafting.
Implications
Combined, the results illustrate two changes in the nature of work:
- First, they indicate a reallocation of time with towards more autonomous rather than collaborative work.
- Second, they also show a transfer of time towards tasks oriented to “manage AI”: model evaluation, editing of first drafts.
AI related skills will be more and more in demand and companies able to effectively train their workforce on AI will get a significant advantage.
Photo de Solen Feyissa sur Unsplash