An MIT study found that only about a fourth of jobs are economically sensible to replace with AI vision systems. | Source: MIT CSAIL
A new study from MIT CSAIL, MIT Sloan, The Productivity Institute, and IBM’s Institute for Business Value provided more insight into how artificial intelligence could affect the job market. Its findings challenge the common belief that AI will put vast numbers of people out of work.
There’s a lot of research that exists about AI in the workplace, but much of it focuses on quantifying all of the potential uses AI can have in various sectors. It doesn’t focus on how likely those industries are to adopt that technology, and what costs might prevent them from doing so. This research examined the economic practicality of using AI for automating tasks in the workplace, with a particular emphasis on computer vision.
The authors claimed that their tripartite analytical model sets this research apart from others. This framework assesses the technical performance requirements for AI systems. Then, it looks at the characteristics of an AI system capable of that performance, plus the economic choice of whether to build and deploy the system.
“As AI continues to advance and reshape industries, we hope that the findings from this study will be a pivotal reference, guiding future explorations and policy-making in the ever-evolving intersection of technology, economics, and the labor market to help navigate the challenges and opportunities presented by the ongoing integration of AI into the workplace,” wrote Neil Thompson, principal investigator at MIT CSAIL and the Initiative on the Digital Economy, in a release.
AI vision systems put a minority of workers at risk, says study
The study found that currently, only about 23% of wages paid for tasks involving vision systems are economically viable for AI automation. In other words, it’s only economically reasonable to replace humans with AI in 23% of the jobs where vision is a key component of work.
“This indicates a more gradual integration of AI into various sectors, contrasting with the often hypothesized rapid AI-driven job displacement,” Thompson said. “We placed our focus on the field of computer vision, an area where cost modeling has seen significant advancements.”
These numbers could change, according to the researchers. If development, deployment, and running costs decline, and the industry transforms to provide AI systems as a service, businesses could adopt AI more quickly, they noted. All of these things lower the cost of investment for companies to deploy AI, making the technology more financially viable. This could lead to more rapid changes in the job market.
A shift to offering AI products as a service, in particular, could quickly change the industry, according to Thompson. This would be similar to how some robotics vendors shifted to offering robots as a service (RaaS), lowering barriers to adoption.
“The implications of this shift are profound: It could democratize access to AI technologies, allowing smaller businesses and organizations to benefit from AI without the need for extensive in-house resources,” Thompson said. “Moreover, this could lead to the emergence of new business models centered around AI services.”
Some jobs may go away, but many will be created with AI adoption
The study also touched on some of the broader effects of AI adoption outside of immediate economic considerations.
With more AI systems in place in the workforce, jobs could open up to maintain those systems, it said. As some jobs are automated, businesses will need more people to manage, maintain, and improve AI and robotics. The team says this can lead to growth at the macroeconomic level, with employment and income growing and living standards improving.
“Broad economic benefits will only be realized when fundamental transformation occurs in how business is done and how workers work,” Fleming said.
Generative AI is still in its early stages, MIT CSAIL says
The team was able to do this kind of research because computer vision systems have been operating for years, providing an abundance of data. The team used this data to assess performance and economic viability.
The data for large language models (LLMs), which power generative AI programs like ChatGPT, is still developing, so there isn’t much that the team can learn about how those systems will affect the job market. The team did say that its research about AI vision models can provide some insight into what the future might hold for the adoption of LLMs.
Brian C. Goehring, associate partner and AI research lead at IBM’s Institute for Business Value, and two members of Thompson’s FutureTech Research Project are also authors on the paper. Affiliate researcher Maja Svanberg and Wensu Li, a post-doctoral student at the Sloan School’s Initiative on the Digital Economy (IDE), also contributed to the paper. Their work was funded by MIT-IBM Watson AI Lab, and is now under review at a journal.
Tell Us What You Think!