Recently I posted an article by Jeff Selingo to my LinkedIn feed entitled “What happens when millions of jobs are lost to automation.” Predictably, those who actually read the article decided it was about education -- specifically, how a lot of people are being educated to do things that no one will be doing after a few years. Meanwhile, people who read only the title commented that even more jobs will be created in our hyperautomated future.
The issue is more complicated. When we defund education and fail to address long-standing social issues, we simply don’t prepare future generations (or even much of the current one) for this future. As you automate away the jobs held by people of color and the poor while cutting access to education, you create a terrible cycle of increasing poverty. Also, in a market economy, automation doesn’t always eliminate a job right away, but it does lower the salary offered for performing it.
It's impolitic to say, but education will not fix all of this. I’ve met some very good, well-educated students who were nearly completely unable to think critically. Human intelligence exists on a continuum, and not everyone is smart enough to be able to participate productively as a “knowledge worker.”
Some of these folks are currently employed at a low wage in various bureaucracies as cogs in a complicated machine. As more of those jobs are automated away, will there be enough bureaucracy to employ low-skilled workers?
Possibly. Look at our industry, which is supposed to be about finding new efficiency. Most IT organizations grow to where they need more people to deal with the people they already have -- or as Oscar Wilde said, “The bureaucracy is expanding to meet the needs of the expanding bureaucracy.” We were supposed to save the world from inefficient nonsense (like the 50 times I filled out my son’s name on forms to participate in soccer), not create our own.
I’ve never liked the A player, B player nonsense. It mostly lets drones from Stanford to hire more clones of themselves while ignoring the rest of the world or acknowledge that luck and a whole lot of money are probably more predictive than an arbitrary nonfalsifiable theory of intellectual caste (that oddly looks a lot more like socioeconomic class most of the time). Also, sports analogies are usually terrible ways of understanding the human condition outside of steroid addiction.
The truth is some people have critical thinking skills, some have to be taught, and some will never really cross the chasm.
Those of you who think we’ll completely automate our way out of needing a working class either don’t really understand where we actually are in AI (too many movies?) or are too isolated from the people who “do” things to understand what they actually “do.”
Recommendations have limits
My local coffee shop has people serving the coffee. The actual making of the coffee is mostly an "art" that could be performed by robots or some kind of dispenser. Can you still charge an insane amount for it? Maybe. (Even the instant coffee at Starbucks is expensive.)
A confusing array of options proliferates any modern coffee shop. The barista serves several purposes a machine can’t. Most of the time I order the same thing, but sometimes I want something different. Sure, maybe the barista makes the same or possibly worse recommendations than an AI with my profile could. But she also helps me get over my discomfort with my selection. Also, if she messes up or if I’m having trouble wading through the options, she’s pleasant, and I don’t get particularly upset. Bartenders serve a similar function -- also, they're often cheaper than therapy.
Don’t get me wrong; a good, strong search engine can serve a multitude of my needs. (Disclosure: I work for Lucidworks, which sells search software.) But sometimes I require a human touch to make a decision.
Driverless Ubers
Uber’s pushing hard into driverless cars. Have you ever used Google Maps when a municipality suddenly started construction somewhere? Has Google Maps ever had a total conniption trying to push you to a route it knew -- while you detoured? Sure, you can add some kind of “construction avoidance,” but ultimately a human needs to troubleshoots this. I suspect at the low end, this will be a big part of how most taxi/Uber/Lyft rides work in the United States: At the high end (black cars) I suspect there will at least be "driver assisted" vehicles, to avoid the hassle of getting out and walking around a mess in the rain.
In less developed countries, driverless cars are probably further away. It's difficult to imagine a driverless car navigating the roads of Pune, Bangalore, Sao Paulo, or Bangkok. Heck, I’m not sure Boston or Rome have traffic controls either.
Landscaping and highly variable topics
Sure, an expert system, computer vision, and a few robots could handle my landscaping most of the time. But due to one-off events and various multifaceted problems, this is a few years away -- possibly further than driverless cars.
My whole yard used to drain into my laundry room. This took several years and different people to resolve. It turned out to entail regrading the backyard, putting in a lot more drainage, and installing a much better drain.
There are a hundred variations on similar problems and many, many variables. Yes, we can create AI systems that can come up with the right solutions ... most of the time. But they will take a long time to develop and train -- and longer still to handle “anything” and make the right choices.
Creative design
Today computers can and do generate logos and brand names, sometimes with considerable success. But more complicated art and design requires a creative touch and insight into how others may perceive it.
It takes a fair amount of empathy to make sure that something looks good and feels good while serving a practical purpose. Robots don’t sit on a couch.
One day maybe neural nets can take over the bits and pieces, but ultimately, creativity is the most difficult intelligence to mimic.
Teaching and authority
I reserve my biggest insults for uncooperative machines -- and frequently disregard whatever it is that they tell me to do when a message pops on the screen.
Projecting machine authority is difficult. RoboCop had to shoot his way through his problems. When it comes to teaching, computer-based training existed long before MOOCs were invented. Not a lot is new here, but understanding how a kid thinks about a problem, then helping them through it requires both empathy and a projection of authority.
Your AI future
Wide-scale automation will require a lot of infrastructure investment to accomplish. A lot of it will happen relatively soon. Yes, smart machines will be able to drive, load trucks, and maybe suggest the caffeine I want. Even coding jobs may be eliminated.
But creativity and that organic “leap” to new ways of thinking will always give artisans a place. There will always be the performance arts. There will always be sticky situations that are hard to pre-program and the people who need to figure out the algorithms or techniques to train machines to do these things.
More jobs? Probably not. But fewer of the old ones may go away than you think.