In this episode, Byron talks about jobs and automation.
At Gigaom.com/quiz, there is a ten question test that scores how likely any given job is to be replaced by automation, be it by a computer or a robot. I’m going to spend ten AI Minutes going through the questions one at a time, and why they’re on the test.
Question #4: How long is the training for the job?
You might score this as 0, which is less likely to be automated, if the training is months or years; that might be a dental hygienist, or an auto mechanic, or a locksmith. On the other hand, you might score this a ten, which is more likely to be automated, for something like a job that requires just a very minimal amount of training, like a custodian. Finally, you could give it a score in the middle if it required a few weeks of training, like an oilfield roughneck, a commercial fishing person, or a flight attendant.
The reason this is an indicator is obvious; the more training required to do the job, the more training the AI must have to do it. Some things can slip by it, though. A custodian may not require a lot of training, but it involves so many tasks that we do naturally, that everybody knows how to do, that it’s spread widely throughout the population, but that doesn’t mean that those are easy to automate. Still, the idea of, the longer it takes to train a job, the more likely it is to be replaced, is an easy test.