As chatbots are deployed, AI whisperers will be employed (rerun)

This episode originally aired on April 19, 2023.

The artificial intelligence boom has created a hot market for a new type of job: “prompt engineer.”

It’s basically a chatbot whisperer, skilled in eliciting the best responses from these tools. They’ll likely be needed in lots of different industries — from tech to insurance.

Marketplace’s Meghan McCarty Carino spoke with Anton Korinek, an economist at the University of Virginia and the Centre for the Governance of AI, who said we’ll all need to acquire some of the prompt engineer’s skills as AI becomes pervasive. An edited transcript of their conversation is below.

Anton Korinek: These generative AI systems, they’re going to really revolutionize how pretty much every company that is engaged in cognitive work is going about their business. So let’s go to the example of the insurance company that you mentioned. I would imagine that within the next few years, a lot of the functions for which an insurance company currently has a call center may be automated using some sort of chatbot that builds on generative AI models. And then you need a prompt engineer, who essentially tells the chatbot how to interact with, you know, the customer buying a policy or the customer who has questions about a policy and so on.

Meghan McCarty Carino: What kinds of skills are needed to do this job?

Korinek: So if you really want to aim for prompt engineer as your job, I think the best preparation is to spend a lot of time working with these models. And you know, one way of thinking about it, it’s a little bit like when you’re working together with another human. Over time, you get to know their quirks, you get to know what they’re best at and how you can best leverage their strengths. And I think a good prompt engineer knows that same kind of stuff about the AI models.

McCarty Carino: Are there any sort of formal trainings that are helpful? I mean, it sounds like it might be helpful to have a basic understanding of a programming language.

Korinek: I would tend to think so. But you know, the funny thing is the language that you’re using for these chatbots, it really, it is natural language. So you don’t need to know any formal programming language. But I guess it helps to know how computers function, and how these AI systems function in broad strokes.

McCarty Carino: Now, the salaries for many of these job postings are quite generous. I’ve seen many sort of well into six figures, above $300,000 a year. Does that surprise you?

Korinek: A little bit, yeah, because it seems like a skill that people should be able to learn relatively quickly. But I guess they need somebody who can do it right now. And right now, there is only a limited set of people who really have exactly that experience. And I think, you know, in the big picture, I don’t know if this is going to be, like, the largest profession that there is out there. There will probably be a few hundred prompt engineers within a year from now, maybe a few thousand. I don’t think this is going to be a job for the masses, except to the extent that all of us are going to be doing a little bit of prompt engineering because all of us are going to use these systems in the not-too-far future.

McCarty Carino: Right. I can imagine that as, you know, this kind of technology gets integrated into a lot of different types of jobs that, you know, the kind of entry-level skill level might become very widespread, and perhaps not a super-sought-after skill set.

Korinek: To some extent, it’s going to be like people expect that you know how to use a word processor when you start a job. And they are also going to expect that you know how to do a certain basic level of prompt engineering or interacting with generative AI models.

McCarty Carino: I think so much attention has been paid to this, you know, role, this new position, because it’s such a classic example of a new technology creating a new kind of job. I mean, how does this fit into the larger conversations that are being had about how AI will affect the labor economy?

Korinek: This is one of those examples where we can see a very clear type of job that essentially hasn’t existed a year ago and that has suddenly been created. I should also say, though, that I am not too optimistic about the impact of this latest generation of AI systems on the labor market. So there’s been this story that there is always some job displacement and some new job creation. We economists, we like to frame it as, is there more demand for labor or less demand for labor? And ultimately, that’s reflected in how much people are receiving in their wages. And I’m afraid that this technology will actually reduce overall demand for labor and that it may lead to wages stagnating or perhaps, in certain skill categories, even declining in the next couple of years.

McCarty Carino: So at this point, does it seem worth it for, you know, individuals who might feel like their work is threatened or, you know, maybe young people who are just starting to think about a job, does it seem worth it to even try to move in the direction of getting the experience to get a job like prompt engineer?

Korinek: I think it’s going to be useful for any cognitive worker to know how to interact with these models and how to leverage them for whatever you are doing. So in that sense, it is useful to familiarize yourself with these technologies because they will help you to do your job more efficiently.

McCarty Carino: But …

Korinek: But don’t count on finding a job as a prompt engineer as your main job title.