On a fine day this past January I found myself at a conference in Switzerland where I became disillusioned to that strategy, and it’s also where I heard my favorite quote of 2017 so far:
“Go build your autonomous cars, but those 10,000 truckers are voters and what we’ve reminded ourselves from Brexit…is that we still live in a democracy” — Anonymous
At that same event I had the opportunity to attend a panel with some leaders from a few household name Fortune 500’s talk about AI, and I left with two very interesting takeaways:
- Why does AI have to be destructive?
- Companies are not going to keep people around to pay for B.S. job titles while robots put them all out of work so where do we go from here…
In relation to point one, I won’t sugar coat it here that it’s a legitimate worry that a lot of people will be automated out. As Andrew Ng recently noted at the Wall Street Journal CIO Conference, AI has the capability today of doing tasks that take one second very well, but that includes strings of tasks that take one second. I challenge you to think about what kind of jobs encompass those actions…
You have some in your mind? Good. If I had to hazard a guess, most of them are not technology enabled and usually entail some sort of manual, laborious, tedious, repetitive, or even dangerous task. Classically, they are the jobs where you pick paper up from one pile and move it to another pile, they are the jobs where you are counting how many pallets or packages came into the warehouse by hand, they are the job functions that people don’t have to be doing because technology can do them for us better than we can do them. This shift empowers people to take the cognitive load they were spending on manual tasks and apply their human capital to something requiring creative and/or social problem solving. These are tasks that AI still has the longest time horizon to resolve, and as Yann LeCun notes it may be 30 years out until we even have the compute power to tackle some of these skills.
To point two, our friends at Oxford identified 170 industries that have a greater than a 90% probability of automation, but out the 702 jobs in their study they also highlight almost equally as many jobs that have less than 10% chance of being automated. They range from everything like Mechanical Engineering at #53 with a 1.1% chance of automation (I guess I made at least one good decision in college after all), but they also identify many non-technical roles such as craft artists (#116, 3.5%), and lodging managers (#12, .39%). So, what I’m trying to say is if you ever wanted to quit your job as a cargo and freight agent (#696, 99%+) and open a Dude Ranch in Montana, now is the time!
Thanks for reading, and see you in the next installment : ).