Only 23% of the wages paid to humans performing vision tasks are economically attractive for AI automation.

Will Artificial Intelligence automate jobs or take jobs out of our hands?

A number of professional organizations have made inferences and predictions in response to this widespread concern.

Goldman Sachs estimates that AI will automate 25% of the entire labor market in the next few years. According to McKinsey, nearly half of all jobs will be powered by AI by 2055. A survey by the University of Pennsylvania, New York University and Princeton University found that ChatGPT alone could affect about 80 percent of jobs. A report by Challenger, Gray & Christmas, an employment agency, suggests that AI is already replacing thousands of workers.

So the threat is imminent?

On January 22, MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) released a new study that attempts to go beyond so-called "task-based" comparisons to assess the feasibility of AI taking on certain roles and the likelihood that companies will actually replace workers with AI technology.

According to the report, the study differs from the traditional cursory approach to the potential impact of AI. Instead, it provides a careful examination of the feasibility of AI in automating specific tasks. What makes this study unique is its tripartite analysis model. The framework not only assesses the technical performance requirements of AI systems, but also delves into the characterization of AI systems with that performance and the economic options of whether to build and deploy such systems.

Contrary to expectations, the new study finds that most of the jobs previously thought to be potentially replaceable by AI are not, in fact, "economically efficient" to automate - at least not yet.

The main takeaway from the study, says study co-author Neil Thompson, a research scientist at MIT's CSAIL, is that the coming disruption of AI may happen more slowly and in a less dramatic fashion than some critics have suggested.

"Like many recent studies, we found that AI has great potential for automating tasks," Thompson said in an interview. "But we were able to show that automation of many of these tasks is not yet attractive."

Early in the study, the researchers gave the example of a baker.

According to the U.S. Bureau of Labor Statistics, bakers spend about 6 percent of their time checking food quality, a task that AI can (and is) automating. A bakery that employs five bakers and generates $48,000 in annual revenue could save $14,000 if it automated its food quality checks. But the study estimates that a simple AI system built from scratch to accomplish this task would cost $165,000 to deploy and $122,840 per year to maintain ...... And that's just the low end of the spectrum.

So, if the study was so ill-considered, why the rush to release it? Could it be that the study's backer, the MIT-IBM Watson Artificial Intelligence Lab, pressured the researchers to come to certain conclusions? (Note: The MIT-IBM Watson Artificial Intelligence Laboratory was created by a $240 million, 10-year endowment from IBM, which has a vested interest in ensuring that artificial intelligence is not threatening.)