Recently, we sat down with our CEO and co-founder Christopher Schrader to discuss the future of work.
Let’s cut to the chase. Many people are fearful and anxious about the impact automation will have on their jobs — should they be worried?
What I find fascinating is that when people envision some dystopian future, they assume that the jobs that will disappear are the blue collar ones… We’re going to get these robots sweeping the streets… But have you seen DARPA’s annual Robotics challenge? This is where the latest humanoid robot technology gets tested in the field and the number one challenge remains opening a door. They can’t grip the handle, these million dollar machines will keel and fall over trying to do so. Is it easier to employ someone 10 dollars per hour to sweep the streets than it is to get a robot to do that job properly — definitely yes.
What about someone who sits in front of a computer the whole day?
Well for those people they sit at a desk, process a lot of data, have a lot of rules to process that data… and all you have to do replace their job is theoretically install a program on their computer… a kind of narrow Artificial Intelligence. You don’t even have to change the infrastructure of their job. You don’t have to build a physical robot you don’t have to do any of that. Jobs that have a lot of data and a lot of rules tend to be generalist jobs, with millions of people working on the same job descriptions everyday… which is different to specialized jobs. If you are a clerk, or banker, or lawyer, or even doctor and your job relies on a lot of rules and data then it is at risk of computerization.
You mentioned Artificial Intelligence, what is AI?
The most powerful machine learning (ML) algorithms today, so called artificial intelligence — it’s really not artificial intelligence — we call it applied, weak or narrow AI, because it relies on a lot of data, a lot of rules, it needs to be trained by a human being in what we call supervised training and can only do a few, very specific tasks. The outcomes of those algorithms can result in observations, analyses that humans can’t predict or see. Findings that no human could have identified. This is particularly evident in deep learning, a technology that relies on stacked neural networks. What is a neural network? A very rudimentary, primitive, mathematical representation of how neurons in the brain work. This is like the intelligence of a flea compared to a primate, that is how simple our models are compared to the brain.
We’ve seen big developments in AI technology in the last year. Recently, professors at Oxford produced a study suggesting 47% of all jobs were at risk of being computerized, what do you think?
This is a common heuristic — we often pick the highest number we hear. So if there was an Oxford study that said 90% of jobs will be replaced by machines, we’d probably be quoting that… it doesn’t mean the study is right. In terms of the situation as we see it, working on ML and AI, we need to train our algorithms with 10,000 images so that they can understand what surprise is in a human face. Now how long does it take a toddler to understand what is a car? Maybe they need to see a car twice, three times. And then they know what every permutation of a car could look like and know to label it a car, or a horse for that matter, or an emotion, so there’s level of intuition or innate learning that far exceeds our most advanced algorithms today.
I suppose asking why computerization of work is a risk is relevant now because, as Peter Thiel says, our expectations for the growth of a technology are linear, but what is actually happening is something more exponential. So we say, hey what can AI do today? And we read the Roomba story about the robot that won’t stop spreading cat feces around a house as it attempts to autonomously vacuum the place — while in our minds we imagine something more like the terminator. We are disappointed. It’s hard to tell when that is going to change. However, there will be an inflection point for these technologies and then suddenly our expectations are that the robot that can barely clean up cat crap while we actually have ‘terminator’. The question is then timeline, and I think the answer is decades if not centuries away. Until we have general AI, which can learn unsupervised. Not being an expert personally, based on our best information today, we aren’t so close to that point yet.
How can we better understand how fast or slow jobs will be computerized?
One way to explain it is through ‘substitution factor’. This is what a lot of people miss out on when thinking about the computerization of work. When our lives are augmented by machines, the way we do work is enhanced by computers… That happens when our job has a substitution factor less than one.
What this means is that a computer program or robot can do a job as well as me 80% or 90% of the time, but there are diminishing returns so it gets harder to replace 100% skills or tasks as you get closer to 100%… You might have a car that can drive itself and can drive as well as a human 99% of the time. But what happens if 1% of the time it skids off the road and kills the passengers? Can you replace a driver with a robot? The answer is no. And that’s going to stay the case for a long time.
But in that time the driver sort of sits there, and as we’ve seen in the Tesla adverts pushes the accelerator and the car mostly drives itself… His job gets easier and indeed the machine does augment his experience as a driver until suddenly he’s out of a job. On a long enough timeline enough of the job will be replaced by a machine where you no longer need the human element. And this will happen in particular to low-skilled, process driven, data heavy work.
Will we see warning signs before our jobs start disappearing?
When asked how he went bankrupt, Mark Twain responded, “well, at first slowly and then suddenly.”
With computerization, initially your job seems better and these machines are helping you out, until suddenly the machine takes your job away. It takes your lunch. So that’s the low skilled area where it’s easy to see how machines will first augment and eventually replace most jobs. On a longer timeline this trend will eventually affect all jobs, not just low-skill ones.
Your company, Pathship, connects companies with a network of experts, who some might say work part-time… What is your view of the gig economy?
My personal view on the gig economy is a bit different and doesn’t apply to Pathship. When we look at successful companies in the gig economy today — I would essentially say these companies operate in the ‘robot’ economy.
If you look at the jobs that are being substituted by gig-driven technology companies like Uber, people in these jobs are working for algorithms, their jobs are process oriented, and they are low-skilled or unskilled. Basically, for better or worse, these people are operating in their jobs like robots. The only agency they have is to turn on the app or to turn off the app.
What we are seeing is a transition from a human capital economy to a computer/robot economy, which first of all is quite scary to think about. But in reality, few people want their kids to grow up being cab drivers so most of us are okay with this transition. It creates a paradox because on the one hand our lives are definitely improving because of technology, but ultimately a lot of our utility as humans will be polished away by better performing robots.
…We read the Roomba story about the robot that won’t stop spreading cat feces around a house as it attempts to autonomously vacuum the place — while in our minds we imagine something more like the terminator.
What about more educated professionals — will white collar workers face the same risks?
First, people need to understand that there are costs associated with developing and training an algorithm. If people get more and more specialized, they will retain value for the very practical reason that no one will spend the millions of dollars required to replace you because you have such a niche skill set and you keep those at the leading edge. No one is ever going to bother designing and implementing those kinds of algorithms, even then, by the time they do, your skills will be a step ahead again. Compared to machines, our differentiator is that we adapt and work well in unstructured environments, which algorithms are terrible at right now.
Second, looking at how companies have improved over time and how their structures have changed, competition exists at the margins. If I am an incredibly specialized worker, or as Charlie Munger says, if I’m an expert generalist, someone who is incredibly skilled in one or two areas, and has general knowledge across hundreds of fields and industries so that they are able to relate specialist skills to broader context… Why would I spend my entire life working on one problem at one company? 90% of my job is probably unrelated to my specialization. After all, these problems come and then they go –you are there to solve them. So if you are incredibly good at what you do and incredibly specialized, it’s likely that a lot of companies have little need for your work… highly valued, but short term need for your work. More and more companies will view people as solutions to problems rather than long-term assets. This is exciting because that 10% of your job you love, could become 100% of your job with computers doing the boring stuff.
So what we’ll see are people with these fringe skills, taking their work to companies for years, or months or even weeks to solve specific problems before moving on to the next company.
Is white collar work at risk of being computerized?
To answer this, we need a little historical context. 20–30 years ago companies had ‘star-athlete’ managers, which means your managers were the most skilled workers and that’s why they were promoted. At a Teal organization or at a modern technology company, actually your most skilled workers are your frontline, grassroots workers. Typically, they are super specialized, have a number a degrees and total mastery of their profession… and their managers can’t do their jobs as well as their direct reports so they become coaches and facilitators rather than star athletes. We are entering an increasingly specialized economy.
I think a lot of the tension between companies and workers to do with the security that comes from insurance, education, benefits, etc… my personal belief is that governments should look to substitute those services, so that the obligation is not on companies and it becomes part of your rights as a citizen of a country. If I have a guaranteed basic income, which always ensures a certain quality of life, healthcare that is guaranteed, and a certain amount of insurance — then I can participate in the economy of tomorrow. Increase safety nets so livelihood doesn’t depend on the company.
What’s your message to people who are nervous about an increasingly computerized economy?
I would just say that it is easy to be anxious and fearful of the future, it is hard to be excited about the new opportunities. When the boring work gets automated it will free us up to focus on the fun and meaningful aspects of our jobs that we enjoy. It’s just hard to be optimistic because you have to be prepared — you have to be reading and continuously educating yourself about these changes, lest you get left behind. I don’t think we are walking into some dystopian version of the future, it’s really exciting to be alive, breathing and working today — jobs are only going to get better, quality of life is going to get better, whether that’s the amount of time we have to work, where we can work from…
To find out more about Pathship and how we can help your company navigate the future of work, click here
