This MIT study says AI won't take my job. But it doesn't make me feel too sure that AI won't take my job.

• A study by MIT says it's unlikely AI can take over certain types of jobs — for now. 
• The study looked at visual processing specifically and found it's cheaper to use humans.
• Still, I'm not so sure that I shouldn't be worried.

According to a study by MIT, it might not be so likely that AI is coming for your job.

Uh, but only in really specific instances. Don't get too comfortable.

The study looked at jobs where AI could be used for some sort of computer vision to replace regular human vision. The example they give in their paper is of a small bakery, where a small portion of the baker's job is to visually scan the flour and eggs before mixing them to make sure they haven't gone bad.

Potentially, the bakery could purchase a costly computer vision software system, but as the study reveals —and any bakery owner with a brain would know — this just doesn't make good sense.

Researchers surveyed workers to figure out how many of their tasks were things that could be performed by computer vision, then made some models about cost effectiveness.

Pretty much across the board, even for big companies like Walmart, they found that it would rarely be worth replacing workers with AI visual detection. From the paper:

We find that the median employee works in a firm where close to none of the vision tasks are cost-effective to automate. Even a firm with 5,000 employees, i.e., larger than 99.9% of firms in the United States, could only cost-effectively automate less than one tenth of their existing vision labor at the current cost structure.

The paper agrees that the costs for AI will come down over time, but the authors believe not as rapidly as some others suggest — and that it would take quite a while before AI made a significant dent in these kinds of jobs.

Here's the bad news: This is only a study for jobs that could be replaced by computer vision, which is a pretty specific thing.

For me, I'm not so worried about computer vision taking over my job writing this article, but uh … well … you know. ????