We’ve been as curious about artificial intelligence as everyone else, so recently we asked ChatGPT to recommend a DC-based AI expert we could interview. The tool quickly generated a promising list, including the Center for Artificial Intelligence and Society at the University of Maryland and the Future of Humanity Institute at GW. But when we dug a little further, our initial optimism disappeared. Those centers exist, more or less, but they’re at USC and Oxford, not local universities. ChatGPT had given us incorrect information.
So we turned to a more old-fashioned method—Google—and soon arrived at a local expert: Alberto Rossi. A professor at Georgetown University’s McDonough School of Business, he runs the somewhat new AI, Analytics and Future of Work Initiative, which studies how artificial intelligence might affect our workplaces. We called him up and kicked things off with a little AI assistance.
We consulted ChatGPT on how to start this interview and it spit out the following: “What do you believe will be the most significant ways in which AI will transform the nature of work in the future?”
Oh, wow. Well, for decades, a lot of emphasis was on trying to understand how much automation is going to substitute blue-collar jobs and what the impact will be. Most of the time, when it came to intellectual jobs, people thought that AI was outside that scope. This is changing. You are starting to see that a lot of creative work can be sped up with AI. Some people who thought a white-collar job was a safe bet have had the rug taken out from under their feet in the last year.
Then again, we’re still in the experimental phase. Most studies I see are trying to understand how these tools affect the productivity of individuals. In many cases, in the context of writing, the consensus seems to be that the tools are improving everyone’s final product and boosting productivity. Something that’s interesting is that the satisfaction of individuals is increasing, not decreasing, when using these tools.
If AI speeds up work, will we have more free time? Or just a bigger workload?
Unfortunately, I don’t think this is going to make life easier in the sense that you’re going to have less work. I think your output is going to increase. It may not be the expectation that you have one or two things written a week, but five or six pieces a week. It’s going to be a constant speeding-up of the process.
Sheesh. Will we at least get a pay raise?
I don’t know the answer to this yet. If the output increases, you would expect to have higher compensation. But it’s not clear how much companies are going to charge for some of these [AI] tools. One strategy companies do is offer products at much discounted prices to get everyone on the platform, and the moment they hook everyone, they increase the price. There is a possibility you’ll be paid more, but there is another possibility where licensing the software will cost so much that your wage won’t be much higher.
“In many cases, jobs won’t disappear—they’ll just change. What people do is going to change.”
Who should be most concerned?
The speed at which this technology is evolving is such that it’s hard to predict. One application that seems natural to me is marketing. I think marketing jobs are going to be incredibly disrupted by these tools. If you’re [a business] and you want to post ads across different platforms, you’re going to have a different message on Instagram, Facebook, in print, on TV, on the radio, because the audience will be slightly different. That requires a team just to cover all the channels. But now, that personalization can be done relatively quickly in a fully automated way.
I recently saw that a Chinese PR firm replaced its copywriters with an AI tool.
I think you’re going to have this honeymoon period where, for some people in these roles, their job is going to become very simple. They’re going to be able to use the tool as a crutch for the first couple of years, but the company will start realizing that they may not need that many people in a certain department.
Since we’re in DC, how will we see politicians and lawyers use these tools?
It’s an interesting question, because politics has become more and more sophisticated when it comes to trying to win elections. Traditionally, people would apply analytics tools to figure out where they should focus and deploy more resources in order to win specific counties or constituencies. Going forward, the messaging is also going to be affected by some of these tools. I can imagine having very personalized messaging and campaigning that’s targeting individuals’ preferences.
Lawyers—their jobs are changing dramatically. You had so many people reading through contracts that were hundreds of thousands of pages and lawsuits that were hundreds of pages. Now some of these bots can at least parse through the relevant information and highlight the critical parts so that they can be evaluated by the individual.
Could a robot represent me in court?
That is not out of the realm of possibilities. Right now, when it comes to robo-advisers or automated help, people are comfortable with an algorithm if the decision they have to make is not very high-stakes. If you’re going to a new city, you’d traditionally go to the concierge of your hotel and ask for restaurant recommendations. Now Open Table is going to recommend the best-fitting restaurant for you, depending on what else you’ve eaten. But with more consequential decisions, like medical decisions and lawyers, I think people will still feel much more comfortable interacting with a human—for a while at least.
A lot of the work I’m doing right now is trying to understand how much people follow or don’t follow, like or don’t like, the advice of an algorithm compared to a human. Of course, this kind of research has been completely thrown off the rails over the past eight months.
In what way?
When it came to interactions between humans and bots, the experience people used to have was pretty bad. If you tried to do anything on a website such as changing your flight, even just a year ago, it wouldn’t work. You would have had an extreme degree of frustration. Now with the advent of ChatGPT, you may actually start seeing the paradigm change. More and more companies are trying to experiment with having a chatbot that is fully automated, without any sort of human interactions. So some of the work I’m doing is trying to understand how much and in what context can you fully automate some of this.
This all makes me wonder how worried we should be about massive unemployment.
It’s funny—I never thought about this question before a year or two ago. Past revolutions, like the Industrial Revolution, were always associated with more employment opportunities. But I don’t know if that is the case anymore. In many cases, jobs won’t disappear—they’ll just change. What people do is going to change. This may require considerable reskilling. Having some sort of subsidies for individuals who may not be employable—that’s something to think about going forward. But the crucial thing is trying to make sure there are opportunities for workers to be reskilled.
What type of jobs might be created by AI?
One of the jobs that’s been created because of ChatGPT is prompt engineering, which is a job I didn’t even know existed until a year ago. It’s similar to the way you figure out the best way to do a Google search. With ChatGPT-type tools, prompt engineering—basically, what kind of query you put in, what information you upload with a query—makes a huge difference in terms of responses.
And certain industries will actually expand. With financial services, people were concerned that the moment you have robo-advisers come in, the demand for financial advisers would disappear. But exactly because the robo-adviser is able to reduce the cost of providing service to people who are less wealthy, you have the market expand dramatically.
In regard to where AI is right now—is this about all it will do? Or are we just at the start and it’ll become far more advanced?
One thing is becoming more and more clear: The more data you feed these tools, the better they become. The development of these tools is going to become faster and faster. The more individuals interact with these tools, the more the tools acquire information. I mean, the adoption of ChatGPT is incredible—faster than any other tool we have seen so far. We’re going to see a major leap forward when ChatGPT 5 comes out, and the progression is going to increase.
I guess I’m still skeptical about these tools. It’s hard not to think of the movie “Idiocracy.” Have you seen it?
The premise is that everyone has become so spoon-fed by technology that they aren’t able to think critically anymore.
I see. People were very scared when books came out because they said, “Well, if you have books, you don’t need to learn anything. You can always read it.” When the internet came out, people said, “Whoa, this is terrible. We won’t need to learn anything or even read anything because we can always just find the information.” But that’s not what turned out to be the case, right?
Having access to Google and so on and so forth has made us much more effective and productive. Now you have this next level. Of course, at the beginning, it’s a little bit difficult to absorb these changes, but I think people will naturally evolve. We’re just at the very beginning of this revolution.
This article appears in the August 2023 issue of Washingtonian.