This is largely because artificial intelligence tools are too costly for companies to deploy on a large scale. It would not be in their financial interest to dispense with certain roles that are deemed suitable for automation.
MIT researchers came to this conclusion after examining the cost-effectiveness of computer vision software – computer programs that extract meaningful information from images, videos and other visual inputs.
The academics found that employees who perform similar tasks are a far better investment for the companies that employ them than the AI tools that are designed to replace them.
“We find that only 23% of the wages being paid to humans for doing vision tasks would be economically attractive to automate with AI. Humans are still the better economic choice for doing these parts of jobs,” one of the study’s co-authors, Neil Thompson, told TechCrunch.
Since the very concept of computer vision may seem rather abstract to the average person, the academics use the example of a baker to illustrate their point.
A baker spends around 6% of their time checking the quality of food, according to data from the US Bureau of Labor Statistics. This aspect of the job could, in theory, easily be automated by AI.
In the case of a bakery employing five people, each with an annual salary of $48,000, automating quality control could save US$14,000 a year.
But, deploying an AI tool to perform this task would cost US$165,000. On top of that, the bakery’s owners would have to pay US$122,840 a year to maintain the computer system. Which just goes to show that this is not a financially astute decision.
While this study has methodological limitations, it does provide some answers as to the potential impact of artificial intelligence on employment.
Several reports, including one by the consulting firm McKinsey, estimate that this technology will wipe out many jobs. This has given rise to a feeling of unease – and even anxiety – among some workers, who fear their skills will be made redundant.
For the time being, however, it seems that unemployment is unlikely to rise significantly as a result of AI, even if there is no doubt that the technology will rapidly spread through companies in all kinds of sectors.