Will New Technologies Complement or Commodify Expertise?
NOMIS Research Project
Despite widespread popular and academic concern that artificial intelligence (AI) and robotics are ushering in a jobless future, the industrialized world is currently awash in jobs. The challenge that workers face is not job quantity but rather job quality. Middle-skilled workers without college degrees are increasingly relegated into low-paid, in-person service work that offers little opportunity for developing skills or increasing income—even as highly paid, highly skilled specialties comprise a growing proportion of all work. Recent AI breakthroughs may intensify or, more plausibly, reshape these secular forces, creating an urgent need to understand the relationship between technology and expertise.
The project Will New Technologies Complement or Commodify Expertise? seeks to understand how emerging innovations—especially AI—could change the demand for labor by increasing the value of expertise (for example, by creating new types of skilled work or extending the reach of existing expertise) or reducing the value of skills (and undermining pay) even if jobs are not actually lost. It is exploring these impacts both for workers who use these technologies and for the labor market as a whole, inclusive of those indirectly affected. Using historical and current data, as well as field experiments in collaboration with the developer of a large language model (LLM) and a large online skills marketplace, the researchers are pursuing four main questions:
1) By how much have automation innovations (substituting for workers’ inputs) accelerated relative to augmentation innovations (complementing workers’ outputs), and for which skill groups?
2) Is this acceleration explained primarily by technological fundamentals, by incentives or by both? Can those incentives be shaped to speed augmentation as well as automation?
3) How do the productivity and employment impacts of augmentation innovations and automation innovations differ?
4) How will generative AI affect the value of expertise at scale?
The project will harness and, in some cases, develop disparate data sources, including representative US Census Bureau data as well as newly digitized historical data and industry-level productivity data. It will also launch an ambitious field experiment studying the use of generative AI for work tasks, performed in conjunction with both an online skills marketplace and the developer of a large language model. This research will inform efforts to shape technology to complement rather than undermine the value of labor, potentially helping to buttress the quality—not simply the quantity—of available jobs.
The Expertise project is led by David Autor at MIT (Massachusetts Institute of Technology) in Cambridge, US.
Ford Professor of Economics
Massachusetts Institute of Technology (MIT)