Will AI Really Take Our Jobs? And If So, Then What? — With Albert Wenger

0:00Let's discuss or debate whether AI will lead to job loss and what to do about it. All that coming up right after this.

0:38Welcome to Big Technology Podcast, a show for cool-headed, nuanced conversation of the tech world and beyond. We have a great show for you today. We're going to talk about whether AI will actually create job loss, whether we need UBI, and what's coming down the pike in that world. We're joined by a terrific guest, someone I got a chance to meet last year, and I'm so happy to bring on the show. Albert Wenger is here with us. He's a partner at Union Square Ventures and the author of The World After Capital. So, Albert, welcome to the show. It's good to be here, Alex. One broad question to begin with, do you believe that the current state of AI is going to displace people from jobs?

1:15And then, I guess, looking down the road, do you think the next wave is really going to start doing that? I don't like the framing of this question. We want automation. And when I say we want, who is the we and why do we want it? We, I think, is humanity. And why do we want it? is because automation has been good to us, right? If we hadn't really brought a lot of machinery to bear on agriculture, we would all still be working in agriculture. It's not that long ago that 80 % of the population was working in agriculture, so that 20 % of the population produced something else. So historically, it's been good for us to use technology to do things that we needed humans to do.

1:57The question is different. The question is, how can we create a society, a social system, an economy that takes the benefits that this creates and distributes them broadly so that people are excited about these prospects as opposed to being fearful? I strongly disagree with you here. And I think this is where we're going to have our point of disagreement, but maybe not from the place that you think. Because, you know, what your answer. So my question was, do you think we're going to see AI job loss? And your answer effectively is yes, and, and this is what we should do about it. And what my counterpoint would be is, you know, maybe we don't see the job loss.

2:40And I don't think we have. I think a lot of companies talk about how they're going to replace people with AI, but they're not. They're weak companies. They're laying people off and they're trying to tell a story to the market about how they're going to use AI for those jobs. And if you, and I think you've spoken about this and we should talk about this, but if you look at how ai is introduced into workflows it doesn't necessarily replace people it might augment what they do and make them more effective give them more time to do things that they are not doing in their day-to-day job but but my point here is is that we've yet to see evidence that this wave of ai is going to take jobs in any large-scale way and i haven't even really seen much practical evidence of it doing it at all yet well first of all and that's a conversation worth having first of all i think i think it sort of depends on what evidence it depends on what evidence you look at right so the ft publishes interesting study of the wages people are getting in freelance marketplaces for copy editing and they looked at it um and since the release of chat gpt there's been a market downturn in wages for people in jobs like copy editing in freelance marketplaces.

3:52I think there's always two separate discussions here, right? There's the discussion of what happens in the long term, you know, and then there's the discussion, what does the transition look like? And I think the transition will be harsh because we're unprepared. And I do think this wave is completely different from the waves that have preceded it. By the way, I would also point out the following.

4:16people were like, oh, automation doesn't seem to have any impact so far. Why are we not seeing it in the data? But oftentimes what happens is that companies can afford a huge amount of bloat because they're doing really well. And it's only in downturns that companies go, oh, we really got to tighten the belt. And so a lot of the stuff that has happened in the past in in terms of automation didn't play itself out during good times. It played itself out during belt tightening times for companies. The other point is, I think this wave is very, very different from the prior waves. And I still think this particular way of framing the discussion, have we seen it, is like, I think labor may be getting too cheap very rapidly, which might then not lead you to a lot of automation.

5:12If you look at what happened in the first go-around with machines, so meaning the industrialization, if you look at countries like Europe or European countries, especially a country like Germany, that had a strong union movement, wages went up quite rapidly, working conditions got better quite rapidly, that required the formation of capital. You could only compete globally or even regionally if you were able to build machines. Then you look at countries that had a complete surplus of labor for a long time and wages stayed super low. There was very little capital formation in those places. So I think there's sort of this weird cause and effect thing that people really have confused here, which is why would I, if I'm McDonald's, put a robot into the kitchen if I can pay somebody minimum wage to do it?

6:09It's just not as incentive. Why would I build a toilet cleaning robot if I can pay somebody minimum wage to go do it? So I think this is what I'm saying. I think we are more likely to be stuck in a low automation trap because we have too many people who have to take jobs that aren't in any way, shape or form the kind of things where people go, oh, this provides meaning and purpose in my life. It's like, this is so I can pay rent. This is so I can feed my children, right? So I think we're much more likely to be stuck there. And then the final point I want to say is this whole augmentation versus replacement debate is also, I think, super misleading.

6:46Because let's say, in fact, that I'm augmented and I can now do whatever, you know, I'm a lawyer and I can read contracts faster because I'm an AI summarizing for me. Like, yes, I am still that lawyer, but I may have, you know, instead of having three young lawyers working with me, I may have one young lawyer working with me, right? So this idea that augmentation doesn't put pressure on the job market is also wrong. So I just think what we have done over time, and this has happened already over the last 20 years, is we've made the wealthy wealthier and the poor poorer. And we've created essentially a precariat.

7:31And companies like Uber have effectively sucked up this precariat into these jobs. And companies that have, you know, job scheduling systems that are like, you have to show up here at this time. I don't care whether your child is sick. I don't care whether you have to go to the doctor, you have to show up or me, I fire you, because guess what? I can hire other people. All of those things are kind of intermingling in weird ways to, I think, give us not the level of automation that we want and to give us weird automation and to give us weird social results. So my point is, I think we're just not set up to get good social results out of this coming wave of automation.

8:11We're just not set up for it at all. Yeah. No, look, I appreciate what you're saying. And we had like a little bit of this discussion privately before. And I was like, okay, we should do this on the podcast. So it's actually really valuable to hear your thoughts. And I appreciate you sharing them here. But let's press forward a little bit more because I think we should talk about this with some more concrete examples. I mean, looking at the industry that you invest in, in technology. So we now have these co-pilots that can work side by side with engineers to help them code faster. Now, first of all, I'm not 100 % sure how effective these are.

8:44I've yet to hear a CEO talk about how it's revolutionized their engineering force or a CTO talk about it. Most of them say, yeah, it's useful. We use it. We have a seat on GitHub Copilot for people and we're figuring it out. But let's say it is effective at making engineers three or four times better. So if you look at it from a perspective of a CEO, are they going to then say, you know, we actually need less engineers? or are they going to say we're now capable to build more and we're in a war with other companies who are building new products and features and services and now we can just build more like no ceo says we are not uh we're building just at the right pace on our roadmap every ceo says we wish we could build our roadmap faster and this might just speed that up so where do you stand on that because that might yeah it's never one or the other it's silly to have a debate over whether one or the other.

9:35But it's also silly to have a debate over whether or not it's going to put pressure on engineering salaries. It is definitely going to put pressure on engineering salaries, like 100%. And it's already doing that. At our Portfolio Company Summit, we had this discussion and people were like, oh, a lot of it had to do with, did the CEO say, I'm going to work with my VP of engineering and we're going to use this technology and we're going to lean into using this technology and we're going to basically say if you're not using it like you're doing it wrong or are they like i'm going to let things run exactly the way they've run before so um so i just think it's silly to think that when you have a machine that can do part of the work at basically the cost of energy, let's call it, that that's not going to put some pressure on what humans can earn doing this.

10:33And so for a while, we had this idea that everybody can just get themselves out of poverty by learning how to code. And for years, I've been saying, no, most code in the future is going to get written by machines. And people were kind of laughing at this. And that's what we're going to get. So I just, both can be true. Companies can be building more and they can also be paying engineers less. And some, by the way, it'll always look weird because they'll pay some engineers even more. Because now you have the sort of hyper productivity of some engineers and those engineers will get paid more than they were paid before.

11:02And, you know, look, there are these studies out there. There's this great study where, you know, when ATM machines first came out and people were like, oh, it's going to, you know, get rid of tellers. Well, it turned out there was an early phase of ATM machines where actually banking became more accessible. So people banked more, people had used more cash, people had more interactions with the bank. Actually, the number of tellers went up. And this study became very famous, and it gets cited frequently by people saying, oh, you know, look, we introduced ATMs and actually number of bank tellers went up.

11:30Guess what? The number of bank tellers eventually went way down. So it's like the early phases of this can look like a total head fake. And I think there will be engineers who will get paid more because they'll be more productive. Yes, that is not the question. The question is not will some engineers be paid more. The question is that, will some companies ship more? The question is, what happens to software engineering as a way of earning a living across the board? And I think there the answer is unequivocal in one direction. Yeah. So you're saying that your portfolio companies are starting to pay their software engineers less because they're using Copilot?

12:09Yeah, absolutely. Some companies have used this as an opportunity to go. One of the interesting things that happens is when you get a new technology like this, a lot of people are like well i'm not using it because the code it produces is worse than what i can do and that's true for a few engineers for the vast bulk of engineers that's not actually true and so um what companies are doing is companies are like okay i can use more remote engineers i can use more code generation and yes tying it back to our earlier conversation there's companies that can't raise right and those companies aren't interested in shipping more features on their roadmap, they're just interested in getting to break even because they can't raise.

12:51They're trying to get to cash flow positive. So yes, they're obviously going to lean into this. And again, the discussion is you always have to look at this as distributions, right? So there'll be some companies that are like, oh my God, this is amazing. I'm getting so much better. I can ship more. I'm in a market where I have great margins. I'm crushing it. So of course, I'm going to do more. But if you zoom out and you look at it, you know, people love to say people are Luddites. But what happened to people who were actually manually operating these things, manually doing these things? Their jobs went away.

13:26So in some ways, the Luddites were right. They weren't right about is technology good for humanity or not. But they were right that it was bad for their specific jobs. Right. Look, I have to stand on the table for this line of questioning. I mean, I understand, you know, in your seat, it might seem silly, but I also think that like for people on the front lines, they are going to want to see what signals there are and what they should be looking at. And it matters very much to them. So I'll be done with it in a moment. And we can. No, we should stay with it. It's good. I'm liking this. So then.

14:02OK, so. All right. Let's keep going then. So explain this one to me. We've had all these, you know, revolutions. I would say that we have more AI in the workplace now. I mean, it's sort of obvious than we ever have before. Yet we're looking at near full employment, basically full employment in the US. So how do you respond to that? Well, there's a couple of things. First of all, LLMs are a very recent thing. And we are early in this, right? I think people are always like, well, look, I'm using CHATGPT every day. Why is it not in the labor statistics yet? But, you know, when I was at MIT, my thesis advisor was a guy named Eric Brynnelson.

14:42And Eric wrote the definitive paper on the effect of IT. Yeah, he's been on the show before. We had a great conversation. Yeah. And so, you know, there was all this thing where people like, well, it's not showing up in the macro statistics. But the macro statistics were the wrong place to look. Okay. And you can be at full employment and yet have terrible social dynamics. The two are not mutually exclusive because most people have to work to live. Right. So eventually they go have to find a job. It's not like they cannot have a job. And so, yes, you can be near full employment and you can also be in a situation at the same time where the income and wealth distribution gets ever further apart.

15:25And that's the place we've been for quite a number of years. Now, in recent, very recently in the economy, there's been an uplift in wages. And some of that has come through raising minimum wages. Some of it has come through some areas that are having relatively high demand and having a hard time finding people. Some of that was because during COVID, when people got stimulus checks, they exited the labor market. But people looking at these short-term things and making long-term extrapolations from them, I just think is really, really bad practice. To think about the long-term, you have to think about the fundamentals and you have to think about first principles, not about short-term.

16:03The second argument that's, I think, also not intellectually honest is saying it's never happened before, so it'll not happen now. It has happened before. Has happened before when we had the transition from the agrarian age to the industrial age. That transition was absolutely brutal. It was horrific. Living conditions in cities were terrible. Working conditions were terrible. Working hours were terrible. There was child labor. It was just bad. It was a bad, bad transition. And so if you believe, as I do, that this technology, digital technology and AI in particular, is bringing a big transition, then right now we're seeing the beginnings of a bad transition.

16:45Again, yes, you can have full employment. And by the way, full employment is measured by people who are seeking jobs. It doesn't include anybody who stopped looking. Right. So we have given ourselves a big leg up on how we've defined that statistic. And then we also don't have much about like, what is it actually doing to people's wealth? Can people afford a house? Can people afford an apartment? Where are people living? Can they afford to eat food that is healthy as opposed to food that makes them obese? Like, there's a lot of factors that go beyond looking at the top level job number. And if we're not willing to do that work, we're going to have the same terrible transition because we're not looking at the conditions that people are actually experiencing.

17:28Okay, so let's do that work. We'll talk a little bit about, well, more than a little bit. We'll talk about what you think needs to be done and how this might play out and your views on UBI, which I find pretty interesting. You're running an experiment. We'll talk about that. All that and more coming up on the other side of this break. Did you know your credit card points and miles can lose value to inflation? credit card companies often reduce the redemption value of your points and miles. Now imagine a credit card with rewards that can grow in value. With the Gemini credit card, you can earn Bitcoin or one of over 50 other cryptos instantly with no annual fee.

18:05Every swipe at the store or gas pump earns you instant rewards deposited straight to your account. Plus sign up now for a$200 Bitcoin bonus to kickstart your rewards. Visit Gemini.com slash card today. Check out the link in the description for more information on rates. Again, if you're looking to invest in Bitcoin but don't know where to start, the Gemini credit card makes it easy. The Gemini credit card is issued by WebBank. In order to qualify for the$200 crypto intro bonus, you must spend$3 ,000 in your first 90 days. Some exclusions apply to instant rewards in which rewards are deposited when the transaction posts.

18:42This content is not investment advice and trading crypto involves risk. The Gemini credit card, cannot be used to make gambling-related purchases.

18:54You're used to hearing my voice on the world bringing you interviews from around the globe. And you hear me reporting environment and climate news. I'm Carolyn Buehler. And I'm Marco Werman. We're now with you hosting The World Together. More global journalism with a fresh new sound. Listen to The World on your local public radio station and wherever you find your podcasts.

19:21And we're back here on Big Technology Podcast with Albert Wenger, who is the partner at Union Square Ventures and the author of The World After Capital. It's a pretty funny discussion that we're having, I think, in one way, because usually the journalist and the venture capitalists are on the other side of these conversations. You're the one that's, or the VC is usually the one who's like, don't worry, everything's going to be fine. And the journalist is like, but, but. I mean, that's why we do these conversations. It's a good inversion. I like it. We like to have the nuanced conversations on the show.

19:53And I think by sort of poking at the arguments, we can sort of get to the truth. So let me ask you this to start our discussion about what to do. Do you feel like any conflict here as a venture capitalist who's investing in this AI technology, knowing that there's going to be disruption in the economy because of it? No, I feel very good about the kind of things that we have invested in at USV over the years. You know, like a big part of our thesis has been about broadening access. So it's taking things that historically were accessible to only a few people or not at all and making them broadly accessible.

20:29And if you look at our learning portfolio, a company like Duolingo, for example, right? Language learning, suddenly, at least the beginnings of it, unless until you get to a certain level of proficiency, you can have it for free. So, I believe that a big part of the USV portfolio and also now the climate portfolio is all about the building blocks of this new world. And it can be a really beautiful and wonderful world, right? I think we've got to get away from this fear-driven narrative. It's the fear of jobless. We have to look at what can the world look like where machines can do a lot of things, where we can have time to take care of family, time to make art and music, time to restore nature, which needs a lot of work.

21:18Like there are all these incredible things. We will never run out of amazing things to do. The question is, one, do you have to do them so that you can live? Like, is it mandatory? Is the social system set up in a way where that's the only way you can survive? And I think, given all the progress we're making, we should allow ourselves to envision a world where that doesn't have to be the case. By the way, the beauty of shifting from the existing totally badly broken social system to something like UBI is that today, people face a very large disincentive from working. If you're on any of the programs, the second you start making money, you get kicked out of the program, which tends to have the effect that you're essentially facing 100 % tax rate.

22:08You're getting some new money in over here from work and you're losing all the benefits. So like, why would you start working? So in our trial in the city of Hudson, where 128 people are getting$500 a month, by the way, that's a small sum, some people might think. But you look at a lot of Americans, they don't have$500 in their bank account. Something like 50 % of Americans don't have$500 of savings. It's an extraordinary mind-boggling situation, right? What has happened? Employment has gone up. More people are working in this cohort. In fact, it's gone up 3x. Why? Because unlike the existing social system, which puts a huge disincentive to work, there's no disincentive to work in a UBI world.

22:47Now, it's not a silver bullet. You can't just introduce UBI without also changing the tax code, without changing other things. like it's not in and of itself some magic thing but the existing social system is badly badly badly broken and one way it's broken is that it's a huge disincentive to work like massive disincentive to work and i find it so funny that a lot of vcs and other rich people complain about taxes and have never looked at what the effective tax situation for a poor person is who's getting some amount of social assistance because that tax situation nobody would work none of those people who complain about their taxes would work if they lost literally their entire benefit the second they started working.

23:27And it's interesting that the employment went up in the experiment that you're running. So if you think about universal basic income, the broad idea is to give a stipend to people and use that as sort of maybe a hedge against some of the technological displacement that we might be seeing in the future. And it's interesting to me because we also ran, effectively a UBI experiment in the US, instead of having technological displacement, we had virus-based displacement. And we did the CARES Act, where we gave a lot of money to people monthly to get them through COVID. And then what followed was the great resignation.

24:06So I'm curious what you think about that, whether this was played out on a much larger scale, and you actually did have people say, I don't want to work anymore. I think it's great. I think the great resignation was absolutely fantastic. and we need more of it, not less of it. We need fewer people wanting to work in shit jobs for shit wages because that will lead to technological innovation. Innovation will come when people are like, no, I'm not doing this for this amount of money. That is a driver of technological innovation that we shouldn't deprive ourselves of. We shouldn't be like, oh, we can just pay somebody a terrible wage and we can have the computer tell them when they need to show up for work.

24:41We can fire them for whatever infraction we deem because we, you know, like, no, like technological innovation will come from something like the Great Resignation. By the way, I think COVID is such a terrible experiment, right, on so many levels. First of all, we made large one-time transfers to people. The history of people being good with one-time transfers is very low, right? It's like a windfall, and a lot of people We'll take the windfall and not spend it wisely. So our pilot, the one that Susan and I have been funding, is$500 a month for five years. It is the longest pilot of any of the pilots out there.

25:23And why is that? To get any sense of what people will really do, you have to give them some serious time on this. Not a, here's money once. Like, here's like a lottery winning. Like, we know from lotteries that people don't do very well with one-time transfers that are like a windfall, essentially. And so all of this is to say, if we want to look at this not through a fear lens, but through a lens of what is this amazing world that we can create, we need to allow ourselves to think broadly, not narrowly. So I'm curious, like from your perspective, from a venture capitalist, like there is an imperative to grow and to sort of work intensely to make something happen.

26:10How do you sort of square that with this UBI world where, you know, people don't necessarily have to work and can pursue other interests? Look, math to me is one of those amazing examples, right? So we've made such extraordinary progress in math that the math you learn when you go to high school is math that was all invented by like 18th year, 1800. And even you go to college and start studying math, like it's really hard to come up to speed to all the math that we know. Now, people don't actually become wealthy as mathematicians unless they wind up going off to Wall Street. You can't patent math.