Today, our guest is Byron Reese. He’s the publisher of Gigaom, he’s also the host of the podcast “Voices in AI” and the author of the book “The Fourth Age”, and we’re excited to have him on the show. Hi, Byron, how are you today?

Hi. I’m doing great! Thanks so much for having me.

Of all those titles which one do you like the best? Which is your favorite one?

I spend most of my time doing Gigaom, I mean, we’re a company that advises and apprises on how to use technology and that is a huge mission, I mean, the one that we can kind of almost never live up to but, it’s an exciting one to me.

Yeah. And it’s a great resource, obviously, I think most people are aware of it and familiar with, especially with this topic we’re talking about of Work Minus Monotony. So, why don’t you lead us into this, obviously, we’re talking about artificial intelligence, about how the world is changing, so what to do you mean by Work Minus Monotony?

I have a thesis and it says that if you can think of any job that someday we’ll build a machine to do, I mean, forget whether we have a machine to do it now, just some jobs that you could imagine a machine doing in the future like, a drone window washer, like you can imagine someday there’ll be a drone flies up and cleans all the windows in the side of the building.

If you take a job that a machine could do, and you ask a human to do that job, there’s a word for that, the word is dehumanizing. The work is by definition dehumanizing because if you say machine can do it, then it doesn’t require passion, creativity, emotion, nothing that makes humans human. In fact, it really is you asking a person to just be a stand-in for a machine until we invent some machine to do it.

I think those sorts of job are the very worst. I think humans are meant for more and so that’s what I mean by monotony. And you can think of it in a macro sense of, you know, the example I gave the drone that cleans windows, but I think you can also think of it in a micro sense–individual tasks that you and I do in our jobs that machines could do. So, even if you have a great human “human job” that requires unique human abilities, there’ll probably be still parts of it that machines could do and that’s what I think of this, the monotony part because it’s a non-human part, it’s a part that is just you physically being a machine, and that’s what I think we’re on the path to getting rid of for everyone.

So, you talked about the difference in a task and a job which we’ve had guests in the show talked about this that, you know, it’s not like AI is coming for your jobs right now. There are certain tasks that will be given over to machines but maybe not jobs as a whole. Can you speak to the difference between those two?

I think that’s an important distinction. Oft-cited study out of Oxford by these two guys, Frey and Osborne that it’s often reported that they came into inclusion that automation can make 47% of jobs. What they really came to a conclusion was that it can eventually do 47% of tasks. It’s unclear that is all that many jobs actually. It’s been pointed out that any technology that augments humans–and that’s what most technology does at its core it increases human productivity. And any technology that increases human productivity doesn’t eliminate the job; it just makes the human more productive in that job.

And you can think of a range of examples of this. There are certainly cases where as we automated, we really did need fewer people and because we often say that for ten thousand years, it took 90% of us to produce our food. And now in this country it takes two or three percent so there’s no getting around, we needed fewer people in agriculture and that technology brought that about. But, in most cases, that’s not what happens.

I’m not the first to site the example of the ATM machines, the automated teller machine, which sure looked like it would eliminate tellers but we actually have more bank tellers now than we had when that came out. And the prevailing idea why that is, is because by lowering the cost of opening a branch, this technology calls banks to open more branches, therefore needing more people.

Likewise, you may have heard that Google Translate can translate a simple document to another language, as well as a human, but the outlook for human translators is skyrocketing. And why would that be? Well, the minute you lower some aspects, you take out the monotony part. Translating a simple email, the minute you lower the cost of that to zero, you open up these new opportunities of doing business in other countries. And then of all a sudden, there’s a contract that needs translating; there’s verbal conversations that needs translating; there’s localization that has done for the local customs in that area.

And so, what we find when certain tasks are automated, more often than not, the world response by consuming more of those tasks and that invariably needs more people. I mean, I believe something very simple because there’s big debate about the effect of technology, on automation on employment. Huge debate, right, like you can’t get away from it. There are people who take think this is awful for workers and that pretty soon they’re gonna be all these workers that can’t compete against the automated machinery, and they’re gonna be obsolete.

And then you have other people who don’t just think that’s a little off, they just think that’s kind of the wrong way to think about it. And I wrote a book where I try to explore all of the options very faithfully, really kind of make the case and think them through. But I’m very much in the camp that says there’s kind of no way these technologies can cause an employment because my simple idea is that automation increases human productivity and that is always a good thing. And if you don’t think it’s a good thing then you might want to advocate for a law that requires people to work with one arm tied behind their back because that would lower productivity and you would need more people, you would create an enormous amount of jobs because you need twice as many people to do anything. But, you’ve also decreased productivity which means you pulled wages way down.

Conversely, technology is like adding a third arm, it increases human productivity for everybody and that will always be good for workers, always. There’s not a scenario that says wow it would be bad if humans are more productive, there’s not a scenario. Artificial Intelligence is a really simple idea. It says let’s study data about the past and you use to make better decisions, it’s always a good thing. If not, then you got to come out against good decisions and say, “No, no, it’s better for people that we make bad uninformed decisions than if we make a decision”. So, I’m incredibly bullish on these technologies with regard to their impact on employment.

Is it a potential to have a category of technology that does not increase human productivity?

No. I think that’s an aftermath to the definition of what technology does. I mean, in a mock percent, I mean, you could say that if you made a bomb that destroyed the planet, the piece of technology that does the increase for productivity, but in a real world kind of examples, that’s what all technology does, is it amplifies what people are able to do. You could move more bricks with a forklift than you can carry on your back. That’s its power, that’s the trick we learned. You see, your body uses about a hundred watts of power and that is the limit of what you are able to do. And if you were dropped in a desert island right now, you would feel the limits of that 100 watts of power like that’s one light bulb’s worth of energy, that’s what you are. But what we learned how to do is, on this electricity and technology to amplify what we are able to do. And in the United States, the average person uses 10,000 watts of power all the time. So, in essence, you have a hundred other of you helping driving your car, you don’t have to push it, doing of all the things that we can do. So, no, I believe the very essence of technology is it amplifies human productivity and it is, in the end, the source of all wealth. If you think about it, I don’t about you but I don’t work harder than my great grandparents dead.

But, I’ve lived a much more lavish than what they ever dreamed of living. And it’s because an hour of my labor is so much more productive than an hour of their labor. I don’t haul up water from the well, I just turn on a tap. And so, we’re rich today versus poverty in the past, not because we just work five times harder, but because we have technology that makes our work five times more effective, and so it’s always a good thing.

So, in someone who lives and works and talks about things that are really on the edge of technology, what are some monotonous tasks that people may not even know about that are on the verge of being automated or we can expect the next five years that really will be kind of out of our range of what we do as humans?

I mean, it may fall into a few categories, think about any tasks that two people would do exactly the same like, data entry. You would have two people do that exactly the same, whereas screenplay writing, it’s something two people won’t do the same.

So, anything that two people would do the same, any tasks that doesn’t vary from day to day is probably automatable. Supposedly, by one measure, any cognitive task that takes a human less than a second, we can probably automate.

The kinds of things we can’t are obviously the converses of all of those. You’ll never see, and ‘never’ is a big word I know, but you’ll never see like a robotic plumber that can come into your home because, you know, your kitchen and your bathroom are different than any other one, your problem is different, the clog is in a different place. Automations are horrible at dexterity, non-repetitive dexterous things like, you know, it can do the same thing kind of over and over and over, but if it has to vary, it can’t do that very well. So, I would say look for things in your own day to day operations that you seem to be doing the same thing kind of over and over always, or you kind of do absentmindedly without thinking and those are probably the kind of things you’ll see automated.

I love your plumber example because it’s something that some people would consider monotonous work, but really it’s extremely varied and extremely, you know, engaging to think about, “okay, how is this think different than another problem”. Would you classify that as almost the opposite of monotony is a work that requires a lot of problem-solving, a lot of thinking?

Yes, absolutely. I think that is the degree to which it kind of changes from day to day. Order taker at fast-food doesn’t change from day to day; it’s probably automatable but somebody who paints street curves, numbers on street curves or something like that actually.

You know, we had this idea, this is kind of funny we had this idea that there are low-skilled humans and high-skilled humans. And I think that kind of misses the boat because any human can do like ten thousand things. They can walk straight, they can balance, they can use a rag to clean up a wet spot in the floor, they can turn pages of a book, they can do ten thousand, a hundred thousand, a million different things, pretty much almost everybody can do.

And then, you take a human who’d does all those million things and they know state planning, and somehow that becomes a high-skilled person, when everybody else doesn’t do estate planning or something isn’t and that I find to be absurd. This is idea that you can kind of automate what a waiter does, but what happens if they send their pizza back because of whatever, or what happens if the baby that the customer brought in needs an ice cube wrapped in a paper towel because they’re teething or any enough for a million other things that that job does. And yet somehow that status is low-skilled job.

So, I don’t believe that there are low-skilled humans. I think everybody is enormously varied and what we can do better than any machine is learn new stuff, and that, by the way, is our singular ability as species, you know. A raccoon can only act like a raccoon, and a snake can only do what a snake can do. But we can study the raccoon and the snake and we can decide, “ah, I’m gonna do what raccoon could do in this case.” We are the ones that can actually learn and do something different than what our instincts drive us to be. I don’t know of many jobs that machines can actually do in their entirety and the few that I do know of, I think will be welcomed to get rid of them. I don’t think anybody is gonna be at any hurry to bring them back.

I read that the ancient pharaohs would take up servants and coat them in honey to keep the flies off the pharaoh. They would go onto the servants. And that to me is kind of the embodiment of the kind of jobs that no humans should do, and there’s a lot of them.

Look, if you enjoy, whatever it is that maybe a machine can do and you enjoy it because it gives you time to think, more power to you. But people I believe, in my experience, when I asked people, “Would you take a slightly harder job for a little more money”? Almost everybody says, “Yes”. And you know I hear this kind of trope over and over which people say this to me. They say, “Look, technology and automation are really good at making new high-skill, high-paying jobs.” Everybody agrees about that. It made some a geneticist, right? And then people say these technologies unfortunately, they destroy low-skilled, low-paying jobs like order taker at fast-food. And everybody agrees with that.

And then, here’s the conclusion that people would always say, they’d say, “Do you really think that order taker is gonna become a geneticist. They don’t have the training for that, how is that order taker gonna get that new job?” And the answer to that conundrum is, “Well, they don’t”.

What happens if the college biology professor becomes a geneticist and a high school biology teacher gets the college job, then the substitute teacher at the high school gets hired on full time to backfill that job all the way down the line? The question isn’t can that person whose mundane job just got destroyed come this new hi-tech whatever? The question is can everybody do a job a little harder than the job they have right now? And I’m bullish on us; I think the answer to that is very much, yes.

And that’s really, by the way, two hundred and fifty years of economic progress in this country. Technologies always made great new jobs, destroyed bad jobs and we all get promotion. Then, it makes new jobs, destroys bad jobs and we all go up another notch. And that’s why we maintain full employment in this country for 250 years in spite of us losing about a half of all the jobs every 50 years.

So, we’re always like burning jobs because technology is making great new ones and automating things that mock these jobs no humans should do. And that’s why we can maintain full employment and rising wages while we’re destroying all these jobs because we just all get to shift up the notch, we all get a promotion.

So, let’s talk about decision making. Decision making is something that automated intelligence is more more capable to do but yeah, we almost like see that as a very high level job of a senior manager has to make a decision, but a lot of times decisions can be codified. So, do you feel like that’s something that would throw a wrench in this linear progression upwards?

Well, machines are only kind of good at a certain narrow kind of decision-making. I mean, artificial intelligence, like I said, is a simple idea. Let’s take a bunch of data, and let’s study it, and let’s make projections of the future. There’s a whole range of problems that don’t lend themselves to that sort of decision-making. Who should you marry? And, I mean, just all of any decision where I don’t have a million data points that I can study to get a definition around it isn’t really–most meaningful decisions can’t actually made by computer. There’s a reason computers do well in games like Chess and Go and it’s because those are constrained environments with set rules and clear winners and losers.

Well, nothing else in life is like that really–a conversation you have, who won that conversation really, how many points did you each get in that conversation, I mean, it doesn’t apply. So, things that look like games, computers can do very well, but that’s a huge minority of all decisions that humans make.

They are, of course, certain jobs that lend themselves to that. I have noticed that when people gave you the same examples every way turn, because I would like to say, you know, radiologist maybe being able to spot tumor or something is data-driven; maybe being a truck driver is data-driven because you’ve heard of those examples used, I assume.

Yeah, because you hear the same examples over and over, generally speaking, that means there’s only like eight examples total and everybody uses the same too because there just aren’t many things that are like that.

I’ve noticed that any headline raised as question, the answer is always no, always, because if the answers weren’t no, it wouldn’t be phrased as a question, you know. Do bananas cause cancer? No, because otherwise the headline would be bananas cause of cancer. And so, it’s one of those things that when I hear the same examples over and over, there just aren’t that many to choose from.

Well, Byron, let’s take this into the context of somebody who’s leading a team in an organization where there’s all these threats going around of, “Now everything is to automation, we’re bringing in robotic process automation, we’re bringing in this and that.” What are some ways that the person leading that team can calm that team down but also be equipping them to make those level ups that they need to make?

Now, that’s a very real-world thing, I mean, I get that, and because you know it’s a broad question that you asked me kind of about all jobs. I would say the simple principles are people should do things only humans can do, and there’s plenty of work for all humans, like I’m looking at a window right now as we’re chatting, and I can see hundreds or thousands of things that need to be done. I mean, there’s plenty of work to be done and machines can do precious little of it.

So, my suggestion to most people is what humans do is we’ll always learn new things so always be learning new things. Look for things in your life that aren’t monotonous; look for ways, if technology multiples what people are able to do, look for ways to use more technology in your life because it’s the only way you increase your own productivity. I mean, you can work smarter and all of that, but the shortcut is, use technology to free up your time and that can be technology in any part of your life.

So, if I were the manager of that team and everybody is, you know, everybody is worried about this, I would talk to my team about what are the aspects of what you do that no machine can do, and let’s double down on those because I need people to do that. And what are things machines could do you shouldn’t be doing. You’re a human being, you’re not a stand in form machine, and you shouldn’t be. And by the way, that’s the lowest value work you can do.

A forklift operator moving bricks is gonna make a lot more money than a person carrying bricks on their back. Why? Because in productivity they can move more bricks and therefore, they can demand a higher salary.

So, that would be my suggestion–focus on the human aspects of jobs, emphasize those, help people automate things the machines can do for them, help people increase their own productivity and if you’re always doing that, you’ll never be obsolete. I mean, never is a long time but you’ll never really be obsoleted if you’re always saying, “How can I use technology to increase my own productivity” over and over and over and over.

You know, when challenge though, candidly, when challenge is how or who reach the financial benefits of that increased productivity? And the way the economy works is kind of interesting. There’s a group of people who sell their time by the hour and have one buyer. Generally, they don’t get the benefits of their increased productivity. If they check out twice as many people as before because they have an automated checking out system, they don’t generally get twice as much pay. Whereas somebody who owns their own time and increases their productivity, like a lawyer. If they can do twice as many wills as before because they have will software, they make twice as much money of, maybe not twice as much, but they make more money.

And so, there are a group of people who sell their labor at fixed costs and don’t get the productivity games that come about. You can think of all kinds of ways to mitigate that. Right now, if you’re a business owner and you have a thousand dollars, the return on capital is probably higher than the return on labor. In other words, you can probably buy technology for that thousand dollars that makes more money than paying employees a thousand dollars in overtime.

But a lot of that is because we directly tax labor. If you hire somebody you pay them salary but you pay taxes on it as well, and they pay taxes. And so, direct taxation of labor decreases the amount of labor that employers buy. Like right now it makes more sense for employers to buy more technology than to buy more labor. But that’s something you can, that’s like a nip and a tuck in a financial system, you know. The structure is still fine but they are policy changes that I think can help people gain a larger portion if their increased productivity.

Yeah. And hopefully, we will make those adjustments as we go through, as we recognize that our economy, and our laws, and governance need to also change along with our approach to work. But that’s a direction that I hope we’re going. Byron, it’s been great to discuss these ideas, getting really deep in these, I think you’ve opened my eyes to a lot of new things. Where can we go to stay in touch with you, obviously, there’s Gigaom, where else can we find you on the internet?

Well, thank you for asking. I’m the easiest guy there is to find. My name is Byron Reese. That’s my handle on Twitter, that’s my Facebook, that’s my domain byronreese.com, all of it like, just type my name and that’s me, the first 800 matches or so. I do write a fair amount and so I invite people who are interested in these topics to read as I write this stuff.

Yeah, definitely check out his latest books as well. It’s a great one. And like you said, it’s a very fair assessment of the world of automation, but it’s also one of the more optimistic ones that are out there, not in the sense of, you know, everything’s rosy and everything, but that takes an honest look and says, “look, we’re gonna be fine, it’s gonna work out and here’s the way we’re gonna do it” so I appreciate that approach you take and thanks for being the guest on the show.

Thank you for inviting me. I’d love to come back.