• As employment and technology experts attempt to forecast the jobs impact of generative artificial intelligence, "AI exposure" has become a go-to way to describe employment threats.
• A recent Rand Corporation study found that jobs which require higher education have the highest levels of "AI exposure."
• As an example, NYU, Princeton and Wharton AI researchers found in their analysis that many teaching positions are among the most "exposed" to AI. 

The rapid advancement of artificial intelligence has raised many concerns, prominent among these AI job displacement and mass layoffs throughout the economy. 

Unlike previous eras of technology and innovation, when job loss fears centered mainly on manual work, this time it is so-called white collar jobs, now also referred to as knowledge worker positions — fields that are cognitively-oriented — seen as being at elevated levels of risk. 

No round of layoffs of knowledge workers goes by without a mention of the threat of AI. When LinkedIn recently laid off nearly 700 workers, many of them engineers, discussion quickly turned to AI replacing a class of highly-valued tech workers — even though, to be clear, there was no indication from LinkedIn that AI was involved in the decision. CNBC reporting indicated that LinkedIn was outsourcing some positions to India.

But there is little doubt AI will be a suspect any time firms are reducing headcount.

"[AI] is going to challenge jobs where cognitive tasks are valued," Alan Guarino, Korn Ferry vice chairman, spoke to CNBC's Andrew Ross Sorkin in a "Squawk Box" interview after the LinkedIn announcement.

"Previously, often, innovation challenged jobs where manual tasks were performed. So, yes, definitely engineers and various others will have to see this as a potential threat for displacement or ideally a tool that will let them do higher order work better," Guarino said.

And that latter point is the crux of what AI and job market forecasters still don't understand — to what extent AI will change the nature of a job that a human still does rather than replace that individual.

In May of this year alone, the U.S. economy lost 4,000 jobs as a result of AI, according to data from employment outsourcing firm Challenger, Gray & Christmas. Given how often C-suite executives are discussing the technology, those numbers are likely to go up. According to Factset, mentions of generative AI on corporate earnings calls increased from five in December 2022 to 390 in June 2023.

A recent Rand Corporation study which looks at technology patents and risk of job replacement found that even before the rise of gen AI, "There are no occupations in the United States that are completely unexposed to general technology patents."

Rand found that to be true as of 1989.

The more recent milestone, according to Rand: by 2020, "nearly all occupations are exposed to AI technology patent."

And the key related point: "In contrast to earlier decades, occupations that require more education and pay higher wages have become more exposed to technology patents in general," Rand concluded.

A recent study from Pew Research Center finds that 19% of all workers in the U.S. are exposed to AI.

In an analysis published earlier this year, researchers from NYU, Princeton and the Wharton School found that among professions "most exposed" to the latest advances in large language models like ChatGPT, eight of the top 10 are teaching positions — telemarketers and sociologists were the others.

"Highly-skilled jobs may be more affected than others," said Manav Raj, co-author and professor at the University of Pennsylvania's Wharton School.

Researchers aren't trying to mask the threat of job losses in softening their talk about AI, using terms like "affected" and exposure. The fact is, it remains unclear whether exposure to AI will be positive or negative across many job functions, especially the further out in time forecasts go. Ask a room of top tech executives what comes next and there is a high degree of optimism about AI at work. At this week's CNBC Technology Executive Council Summit in New York City, 75% of those in attendance indicated in a survey that AI will create more jobs in the next year than it destroys.

During a panel discussion on AI and the workforce at the TEC Summit, Diogo Rau, Eli Lilly & Co. chief information and digital officer, recounted a challenge he made to the Lilly team to come up with any job at the pharmaceutical company that would not be helped by AI. "Still haven't come up with one," he said.

"Every job will benefit from AI," Rau said.

But he made an important distinction in discussing this view, adding that AI benefit to all jobs doesn't mean humans are always going to be the ones doing the job. There will be increasing cost/benefit analysis of whether a position should be human onshore, human offshore, or digital, and some jobs that were previously distinct roles — for example, account management for creative campaigns — could effectively become folded into other positions as AI accomplishes more of the creative support task load.

For now, Rau said, the most important aspect of AI upskilling is just having workers use AI tools. Too many people have opinions about AI without any experience using it, he said. There is considerable work to be done along these lines, on the part of corporate management offering AI training opportunities and being more transparent about how they are using AI — just over half of employees say they have no idea how their companies are using AI today. And workers being motivated to experiment more with the technology.

"Anything that makes us smarter is clearly going to enable us to save time in certain ways, right, so there are huge efficiencies to come here, and those efficiencies will definitely change jobs. There's going to be a reshuffling of who does what and when, and I think the challenge for society is trying to make sure that we make that transition in a smooth as possible way," said Mustafa Suleyman, co-founder of Google's DeepMind, in an interview with CNBC's "Closing Bell" earlier this year. 

AI has the potential to make all employees more efficient – it can help write an email, job posting, marketing campaign, product copy, organize a list, do research, etc. Suleyman, who is now CEO of Inflection AI, which focuses on the development of personal AI tools, says that every individual will have their own AI "chief of staff" within five years.

But as AI gets smarter and more capable, the potential for displacement, or at least a fundamental workforce restructuring, also increases.  

"The net of it is we're going to have to figure out a different way to develop people whose jobs may be displaced at lower levels of the workflow," Guarino said. "For example, upskilling those junior coders faster through formalized training, replacing what used to be on the job training." 

Guarino said upskilling and retraining may not only offer some protection from AI displacement, but could also be positive for the economy. "Over a period of time, yes, we're going to have to find ways to upskill people so they can do higher order work faster and quite frankly, that does all kinds of things, like promote economic growth, efficiency," he said. 

Mario Harik, CEO of logistics company XPO — who worked on AI neural nets as far back as the 1990s —said at the CNBC TEC Summit that it's not only important to ensure workers are getting the proper training to utilize AI tools, but to reinforce the connection between these tools helping the business become more efficient and what that means for jobs.

"You need to make sure they are gaining new skills and are becoming more efficient by leveraging the technology as opposed to being afraid of the technology," Harik said. "But it's also about growing more efficiently, and if you're adding efficiency at a long enough arc of time, the idea is that you can grow by being more efficient and grow your workforce."