So long, prompt engineers

You’ve seen the headlines. “How to Get a Six-Figure Job as an AI Prompt Engineer.” “AI Prompt Engineers Earn $300k Salaries.” “How to Cure All of Your Problems and Worries by Becoming an AI Prompt Engineer.” OK, that one was wishful thinking—but you get the idea.

Over the last few months, as AI hype has approached something of a fever pitch, breathless article after article has touted the same “weird” new job: AI prompt engineer. These so-called AI whisperers are skilled at writing queries for AI tools like ChatGPT that yield effective results and better train the AI.

The hype is catching on: In May, The World Economic Forum dubbed prompt engineering the No. 1 new job of 2023. While some workers have already cashed in—and hats off to them—finance professionals looking for a glimpse into where to invest in the AI future should probably look elsewhere, according to some AI and finance experts only a month later.

Here to stay? “This is something that’s going to go away pretty quickly. In the very near future, saying you’re a really good prompt engineer is going to be like saying, ‘I’m really good at googling,’” Glenn Hopper, CFO and director at Eventus Advisory Group and author of Deep Finance: Corporate Finance in the Information Age, told CFO Brew. “That’s going to be table stakes.”

He’s not alone in that outlook. Victor Sanh, lead research scientist at the open-source AI platform Hugging Face, a researcher who focuses on prompt engineers, also thinks the heyday could be coming to a close. Sanh told the tech blog Mashable prompt engineers are basically like “the PR department preparing someone for a media interview.” They guide the system so it can “behave properly,” but, eventually, those systems are designed to become less fragile and easier to communicate with.

Change happens. Yet even if the six-figure prompt engineer job openings are likely to dry up soon, remnants from this year’s gold rush will still likely change job descriptions in finance down the road, Hopper said. Entry-level finance jobs could be the first to change.

“You’re not going to be an AP clerk or an AR clerk,” he explained. “If 80% of accounts receivable is automated, then the value that you’re going to add is the 20% that takes a human touch to work with clients, to negotiate payment plans.”

A similar shift could also occur across other finance jobs. “If your big differentiator is being really good at Excel—which all finance and accounting people love; we love to show you our formulas and all the cool stuff we can do in Excel—if that gets taken away by automation and AI…then what is in your job description just becomes how you’re adding value to that data,” Hopper explained.

“If anybody can run the report and get the data, where do you get the information from the data?” he continued. “Where do you add value along the line?” That’s the question that CFOs will have to chew over as they determine which AI skills matter most.

Specialize, friends. So, where should AI-curious finance professionals focus their time and energy, particularly as prompt engineers recede into the background? Hopper sees hyper-specific AI models as the most compelling development on the horizon.

“The big players, the big models that are publicly out there, are a great gateway introduction to AI for a lot of people,” he explained. “But what’s really going to happen once the technology becomes mainstream and less of a gimmick is, you’re going to see fine-tuned models, and industry-specific models, and job-specific models that become part of our tech stack that we’re using in our work and our personal lives.”

These new models might come from the same players that currently dominate the landscape, like OpenAI and Google, or from a totally different source. Wherever they start to pop up, though, the generalist model is on the outs, in Hopper’s eyes.

“It’s just going to be a massive productivity boost on all the stuff we’re using,” he said. “I would be looking for purpose-built tools.”