If it ain’t broke, don’t fix it. A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has some welcome news for workers and professionals
It’s simply too darn expensive to implement AI systems that replace jobs.
The study analyzed computer vision - a branch of AI that can take information from images and videos to perform tasks - and found that when you take the high upfront costs into account, only 23% of human tasks can be replaced by AI in a way that makes any financial sense. Most organizations simply don’t have any incentive to throw money at fancy AI systems when their employees can get the job done for less.
Unlike most studies about AI and automation which only look at technical aspects, this specific piece of research from MIT also looked at the economic feasibility of implementing the technology.
The study’s findings stand in sharp contrast to another study released by the IMF last week which claims that almost 40% of jobs are at risk of being automated by AI.
A major caveat to the MIT study’s findings is that a potential future reduction in the costs of training and operating AI models could tip the cost-benefit scale in favor of automation. But the study is fairly confident that won’t happen anytime soon. According to the researchers: “…even with rapid decreases in cost of 20% per year, it would still take decades for computer vision tasks to become economically efficient for firms.”
This article was originally posted by: MIT