Why AI Actually Won't Take Your Job

**Nathaniel Whittemore** (0:00)
Today on the AI Daily Brief, we're discussing why AI actually won't take your job. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.
All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Robots and Pencils, Blitzi and AIUC. To get an ad-free version of the show, go to patreon.com/aidailybrief, or you can subscribe on Apple Podcasts. To learn about sponsoring the show, sign up for our newsletter, or anything else in the ecosystem, go to aidailybrief.ai.
Today, it is a weekend day, which means, of course, that this is a big think episode. And today we're taking on a topic that is just about as fraught as anything in artificial intelligence. That is, of course, the question of job displacement. Every day, there is some new story about a company reducing its workforce, blaming AI at least in part, or some study, which shows all the jobs that could be replaced by AI. And it's not like Americans are particularly comfortable with the state of the economy already. Now, I want to make clear that my argument this episode is not that we shouldn't be concerned at all about jobs. My argument is that in general, we're having the wrong conversations about it. So, let's talk about a few reasons why will AI replace all the jobs is the wrong question. The first problem is that it sort of acts as though white-collar jobs are the only category that matters. Now, white-collar jobs are a big part of the total US workforce, and it is absolutely true that one of the reasons that this particular wave of technology-driven job displacement is hitting people so much differently. Frankly, most of the previous tech disruptions that we've experienced, or that we've had in our history, have hit blue-collar and physical jobs first. The fact that AI is, on the other hand, coming first for white-collar jobs is a real reversal of that trend, with some fairly big implications. White-collar workers are proportionally more economically well-off, and by extension politically enfranchised, which to be clear is not me saying that that's a good thing, it's just the way that it is, and so the potential for backlash, and politically potent backlash to AI, I think goes up. And yet still it's very clear that one of the things that's happening with AI is that it's reminding people that white-collar knowledge-work-type jobs aren't all that's out there. In fact, it's kind of exposing a little bit at least that the pipeline to white-collar jobs, do really well in high school, get into a good college, go massively into debt for your college degree, make it all back with your nice white-collar job, was broken before AI ever came along. College is too expensive and didn't translate well enough into high-earning jobs in general to justify its quality. It was a system that was already on the fritz, even if AI ends up being the straw that breaks the camel's back. You can see already in the early days of AI that its impact in knowledge work is causing people to reevaluate the very fundamentals of which types of jobs they aspire to. That process could lead us into different places where the desiccation of certain categories of white-collar jobs, while impactful is also mitigated by the shifts that people are making. Now, the next reason that we shouldn't be asking, will AI replace all the jobs, is that the very question itself is massively over-rotating on recent announcements. There of course have been a string of layoff announcements where AI was cited as, if not the cause, at least some part of the cause. We had Block laying off 40% of its workforce, big layoffs at companies like Amazon, and more broadly warnings from CEOs around how AI was likely to change the composition of job structures within their companies over time. We've even got scoreboards now of how many jobs have been displaced from AI. And yet, there is also some fairly good evidence that AI is at this point being disproportionately blamed, let's say, for job cuts and layoffs that would have happened anyway. A recent resume.org survey of a thousand hiring managers found nearly 60% said that they emphasize AI's role in layoffs because it is, quote, viewed more favorably by stakeholders than saying layoffs or hiring freezes are driven by financial constraints. Meanwhile, only 9% of those respondents said that AI had fully replaced any roles. A Bloomberg Opinion piece reads, The reason it works is well understood. Decades of research on how markets react to layoff announcements have established a consistent pattern. Investors punish companies that frame cuts as a response to problems. But when a company frames the same cuts as proactive restructuring, the penalty disappears. The stated reason for the layoff matters more than the fact of the layoff. AI has become the most powerful proactive frame available, where restructuring around AI is a growth signal. We over-hired during the pandemic and revenue softened is an accountability signal. So one thing we have to at least consider as we think about how much we should be concerned about AI-related job displacement is the extent to which we are experiencing a wave of AI washing right now. A third reason why will AI replace all the jobs is the wrong question to be asking even though it's being asked a lot is that there are some who think that we're making some incorrect assumptions about how translatable it is to go from AI's disruption of coding and software engineering to AI's disruption of other types of knowledge work. Recently, Carnegie Mellon and Stanford University released a joint study called How Well Does Agent Development Reflect Real-World Work? The abstract reads, AI agents are increasingly developed and evaluated on benchmarks relevant to human work, yet it remains unclear how representative these benchmark efforts are of the labor market as a whole. The study, they say, reveals substantial mismatches between agent development that tends to be programming-centric and the categories in which human labor and economic value are concentrated. Professor Ethan Malik puts it like this, all of the effort is going into benchmarking for coding, but that is a small part of the actual jobs people do, which leaves the true trajectory of AI progress less clear. It is no secret at this point that the entire structure of software engineering has changed because of AI. This is the big disruption that we've been living through for the past few months. What's more, we're seeing coding start to impact how knowledge work works in other roles. When everyone can use software to solve their problems, it's going to change the nature of other jobs as well. And yet, there's a pretty clear through line that some are assuming, from AI can do coding super well to AI can do everything else super well, and there's an argument that particular attributes of coding, for example, its ability to have deterministic correctness and a clear right and wrong, don't actually apply to other areas of knowledge work, which are much more messy and confused and don't have quite the same ability to distinguish correct from incorrect and good from bad. Now, this is a big debate right now, but it's more evidence in the column that all of this is going to be more nuanced than we perhaps think, sitting from our seat watching how AI has just cleaved through the old traditional software engineering process. A fourth reason that will AI replace all the jobs is the wrong question is the extent to which it discounts human preference as a market force. I wrote about this when I was stuck in Brazil as part of my 55-hour trip back from South America with the reflection that although AI was useful throughout parts of that experience, at basically every critical juncture, I was looking for access to an actual human. The reason that I was looking for access to a human is that I didn't want to be subject to the policies as they were written. I wanted to try to talk my way into special treatment. The argument that I was making is that human systems are designed with some amount of discretionary non-compliance built in. If we turn everything over to AI with no ability to use human judgment to make human exceptions, I think systems in general get more brittle. Even beyond that though, a lot of the discourse around jobs assumes that the only function of markets is to be as efficient as possible. But that's means and not end. What markets are actually trying to do is service human desires and human needs. And to the extent that human desires are for other human-mediated experiences, it doesn't matter if everything can be more efficient because of AI, markets will organize themselves around provisioning what people want, namely other humans. Again, this is not to say that there won't be AI disruption. But how much it is and in what areas is subject to a lot of forces that aren't just the onslaught of efficiency and productivity. A fifth reason why will AI replace all the jobs is the wrong question, is that at no point in history has this fear ever been right, at least not in the way that people felt it. There are infinite examples of this, from the Luddites and textile automation to mechanized agriculture, to ATMs and bank tellers, to spreadsheets and accountants, to the Internet and retail. In each case, people spotted the destruction in creative destruction before they saw the creation. But in each case, these were massively market expansionary forces. Now the way things have happened in the past does not guarantee that they will happen the same way in the future. But certainly the pattern that the fear of technological job apocalypse has never actually played out the way that people feared is worth at least keeping in mind when we're considering how much to fear job displacement in this AI context. A seventh reason that will AI replace all the jobs is the wrong question is the one that is the anchor of my ultimate optimism, which in short is the fact that capitalism is radically expansionary. I think that the human capacity for stuff of every type, experiences, services, things is basically unlimited. We are voracious, ever expanding demand machines. I think there's even an argument to be made that the market purpose of technology is to expand the capability of markets to meet this unlimited demand. Joshua Bach put it a different way. He writes, Many people believe that there is a fixed amount of work in the world, and if we give these jobs to machines, humans will not have jobs or starve. This intuitive model of economics is fundamentally wrong. Our wealth depends on the amount and quality of goods and services we can produce and distribute among each other. Automation allows us to make more of everything for everyone. There is always more to do for us. Things we could not afford to do before automation allowed us to get away from the important drudgery of agriculture, manufacturing. And now, documenting, calculating, evaluating, memorizing, and so on. Proof positive of this to me is that 90 plus percent of my AI use cases are not doing stuff I used to do a little bit more efficiently. It's doing new things that I never could be for. And the net result of me doing those new things is not extra saved hours and the same amount of things delivered to all of you, my audience. It's a massive expansion in what I am delivering to my audience. There's a reason that in addition to the podcast, there's a claw camp and an enterprise claw and a super intelligent and an AIDB intelligence and an agent madness. These are things that would not be possible if it weren't for AI. There's also a competitive dimension to this. When Jim Cramer recently asked NVIDIA CEO Jensen Huang why companies are laying people off if AI is supposed to make everyone more productive, Jensen responded, for companies with imagination, you will do more with more. For companies where the leadership is just out of ideas, they have nothing else to do. They have no reason to imagine greater than they are. When they have more capability, they don't do more. I've referred to this in the past as the difference between efficiency AI and opportunity AI. I think it's inevitable that we go through a phase where people are focused on doing the same with less. That's efficiency AI. I also think that it is completely inevitable that because of the nature of our expansionary capitalist system, the companies that win in the long term will be those who opt not for efficiency AI doing the same with less, but for opportunity AI, in other words, doing more with the same or doing way, way more with just a little more. I recently put it a little more crassly on Twitter writing, call me crazy, but I think the companies that give everyone on their team, a team of agents, are going to kick the crap out of the companies that replace their teams with a team of agents. This is one of my most fundamental beliefs, and why I believe in the long run, AI will cause a mass expansion of jobs and opportunity and just overall market size. One final reason why will AI replace all the jobs is kind of the wrong question. Is if that did come to pass, if we all of a sudden saw 15 or 20 or 30% unemployment, we're going to need some totally different structuring of society that doesn't require jobs to be a full participant anyway. In other words, there's no world in which AI replaces all the jobs where society structurally punishes people without jobs. Now, given that this is all a theoretical conversation about what could happen in the future, there are only little drips and drabs of evidence of what that type of societal conversation might look like. But you're starting to see it. Congressman Ro Khanna recently called for a new tech social contract and laid out seven big principles. That same week, Pete Buttigieg talked about the idea of needing a new social contract. If the prognostications of AI job displacement do come to pass, this is the type of conversation that we're going to have. AI replacing all the jobs, in other words, would not happen in a vacuum.