The AI Daily Brief: Artificial Intelligence News and Analysis - How Worried About AI Should Professionals Be?

Episode Date: May 4, 2024

A reading and discussion inspired by https://www.washingtonpost.com/opinions/2024/04/29/ai-professional-class-low-skill-jobs/ ** ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI Breakdown, we're talking about the real impact of AI on the professional class. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.network for more information about our YouTube, our newsletter, and our Discord. Hello, friends. It is, of course, the weekend again, which means it's time for an AI long read. Today, we actually have two pieces that we are going to pair together and talk about on one of my favorite topics, which is what's going to happen to the professional classes when it comes to AI long read. We start with an editorial from Megan McArdle from The Washington Post. The piece is called AI is coming for the professional class.

Expect outrage and fear. She writes, I was born in 1973, just as the American economy was going to hell, although I cannot accept all the blame. It was the twilight of the 20th century manufacturing boom that had almost managed to compress the whole country into one vast middle class, and for reasons beyond my control, that boom was unraveling.

Inflation was headed into double digits as labor productivity began to decline. Economic growth swooned, unemployment rose, and manufacturing employment tipped into the final descent toward its current sub-10% share. This turnaround devastated workers and communities, and even college-bound kids like me absorb the sadness, through songs such as Bruce Springsteins born in the USA and Billy Joel's Allentown, and through shows such as Roseanne. As we entered the electorate, it became a major force in our politics, as the Clintons tried to steer the economy toward a global post-industrial future. Voting for Bill Clinton, I believed his answers were correct. Trade and automation made

Americans better off overall, even as they displaced some manufacturing workers. What we needed to do was to finesse the adjustment, primarily by sending more kids to college to capitalize on the growing wage premium. 40% when I was born, 60% when I graduated college, and closer to 80% today. Older workers who weren't ready to become college freshmen could be retrained for booming service sector jobs. I still think our prescriptions were broadly correct. But as artificial intelligence starts coming for our jobs, I wonder how well the professional class will take its own medicine. Will we gracefully transition to lower skilled service work, as we urge manufacturing workers to do? Or will we fight like hell to retain what we have for our children

as well as ourselves. For I suspect AI is coming for a lot of professional class jobs, despite how many people I hear say a machine can never do what they do. We're accustomed to thinking of automation is primarily displacing the working class, but as economist Darren Esamoglu wrote in 2002, the idea that technological advances favor more skilled workers is a 20th century phenomenon. In the 19th century, steam-driven machines replaced a lot of skilled artisans, and AI looks to be pointed in a similar direction. If you work with words and symbols, AI can already do a surprising amount of what you do, and it is improving with terrifying speed. As a Bloomberg News headline put it in February, AI is driving more layoffs than companies want to admit.

And though the numbers aren't enormous, Bloomberg cites one source that found 4,600 AI-related layoffs during the previous nine months, that's a pretty big number considering that chat GPT was released to the public only in November 2022. It's going to get bigger still. As with previous rounds of automation, good jobs will be created as well as destroyed. But even those who don't get them will enjoy broadly rising prosperity. But that was also true for manufacturing workers displaced by the China shock, as people like me kept telling them. A report last year from the Congressional Budget Office notes that though incomes increase most among households in the highest quintile between 1979 and 2020, average incomes increased in all quintiles.

Yet as they keep telling us, people don't care only about their role as consumers. They care about their role as producers and more broadly about their relative place in society. For the working class, that place has been eroding in relative terms for decades. The kinds of jobs many of them now occupy, in retail or on the lower rungs of the health care system, have less social status than the old manufacturing jobs, even when they pay as well. and they often require a combination of servility and soft skills that wasn't demanded on an assembly line. Those people might be happier if AI improves their relative position, taking over an increasing share of high-status high-paid knowledge work, while leaving humans the lower-skill tasks it still

struggles, such as chopping vegetables or helping an elderly person use the bathroom. In theory, perhaps we should all be happier, as this would be what political leaders have long claim they wanted. A return to the mid-century paradise when the college wage premium was modest, opportunity was more broadly distributed, and incomes were compressed into a narrow band. But if so, we are also likely to see a revolt of the educated people who are losing ground, similar to the revolt that led the working class to embrace protectionism and Donald Trump. Or at least, that's how it seems to me, when I try to imagine the upper middle class

offering their own kids the advice they've so liberally dispensed to working class men. I'm sorry, but the jobs your parents had aren't going to be around, and it's time to face reality and look for steady work in food service or a warehouse. I'd be fine with that. Some educated parent will inevitably write me. As long as they have good health insurance and a strong social safety net. I applaud those public-minded people, but realistically, I doubt they're the majority. upper-middle-class families, I expect there will be a lot of outrage and fear, and demands that the

government do something to help them maintain their position and pass what they have onto their children. Since the knowledge workers are a lot closer to the centers of power than the manufacturing workers were, I expect the twilight of the elites will feature even fiercer and more destructive political battles than the ones we are currently enduring, a Ragnarok of the reasoning class, if you will, an Armageddon of the academically inclined. So I will say first that none of the concerns she has are unreasonable. These are totally legitimate questions to be asking, and And as is clear to anyone who's interacted with many of these AI tools, many of the things that people spend their time getting paid to do now, AI will be able to do at a sufficiently similar

level that it will, if not replace them, certainly impact what they spend their time on. I also think that McCartle is right that there are going to be political implications and political battles. Powerful tend not to give up power. And so to the extent that there is, call it white collar displacement, there will be significant battles around that. Why, I think there's more room for optimism than perhaps columns like this tend to give credit for, is that they always look at the set of activities, the composition of activities that people are doing right now, as somehow the sum total of the activities that people will always do. What I mean by that is they think that there are X number of writing jobs and X number of art jobs, and so if AI makes their 50% of X of

those jobs, then that means 50% of those people are out of work. The math is compelling, but it's inaccurate. The reality is that our market economy is extremely good at using people's time. and filling in whatever excess capacity they have, especially at these skilled levels, to create more stuff, to provide more services, to make more things. It's hard to imagine from where we sit now, because none of us live in the future. But there will be things that we consume in not that long from now, five years, 10 years, certainly, that we literally can't imagine consuming now. Highly personalized entertainment content, as one example.

Will all of a sudden, all that demand for the creatives that were displaced by mid-jurney come roaring back, when production studios want to create something, certainly powered by AI, that needs to be able to output a thousand different versions of itself for a thousand different types of preferences. This also seems totally possible. Why I think it's worth still spending time on these questions even though I'm optimistic about what happens on the other side of this, is that the liminal period in between this composition of the economy changing very significantly and the future shape of the economy taking form, real people will be affected and experience loss and change. It's worth asking, are there ways to make that transition work better? It's not a coincidence

that I'm spending all my time creating educational resources that live in that middle space. There is also one other thing that's worth noting, which is that it is really just not yet clear to us whether ultimately the biggest impact of AI in most professional settings is going to be to eliminate entire categories of jobs or simply change the composition of tasks that most jobs do. Right now, there is certainly good evidence that the way that AI is infiltrating the enterprise in the workplace is not some big top-down hacking off of entire categories of work, but instead bottoms-up adoption, where people are doing things either a little bit more efficiently,

saving themselves time or a little bit better, creating a better output. It's a lot of tiny wins that aggregate into bigger wins, and a lot of the benefit is experienced individually, not just on the level of the enterprise that that individual is contributing to. I think this is also optimistic, and I think that this bottoms-up adoption is likely to be what we mostly experience. I also think that there's going to be a rediscovery,

of what really matters about humans in a lot of different roles. I'm not going to read this entire additional piece, but it's a great example of this. It's by Parmi Olson, and it's called AI Can't Reject Your No Good Very Bad Idea. Basically the gist of it is that as every marketing agency tries to, into AI-based creation, one thing that gets missed is the taste-making of agencies who have to tell their clients that their bad ideas are bad. Said one animator, what we give to clients is the ability to say no to their ideas. They're not visual people, and they know what they think they want.

and then a lot of times it really needs tweaking, sometimes in a major way. If you have ever been on the agency side of a relationship with a big brand, you will probably resonate with that statement. The piece goes on to argue that AI ultimately will just be sycophantic. It will give a brand exactly what it wants without any consideration of whether it's good. To determine whether a thing is good, it's going to take people. And if you don't think this is valuable, think about the difference between great TV ads and the vast majority, which are terrible.

The difference is based on human taste and the ability to execute creative ideas. I think that taste and discernment are likely to see something of a renaissance in a world where the sheer amount of stuff, and especially digital content just goes up as much as it's likely to go up. We will want even more people to cut through the noise and share what actually might get our hearts thumping. Now, how this all plays out in terms of the structure of agencies or marketing or in the entertainment space. Who knows? It's yet to be seen. But it's yet one more reminder that change is neither good nor bad. It is simply change. It is scary, of course, but a lot of how it ends up is down to how we handle it.

And I think there's a lot more opportunity in AI than many people might think. Anyways, guys, that is going to do it for today's AI breakdown LRS. Thanks, as always, for listening or watching. Until next time, peace.