A new generation of artificial intelligence is poised to turn old assumptions about technology on their head.

For years, people working in warehouses or fast food restaurants worried that automation could eliminate their jobs. But new research suggests that generative AI – the kind used in chatbots such as OpenAI’s ChatGPT – will have its biggest impact on white-collar workers with high-paying jobs in industries such as banking and tech.

A report published on Feb 1 by the Burning Glass Institute, a non-profit research centre, and SHRM, formerly the Society for Human Resource Management, stops short of saying the technology will do away with large numbers of jobs. But it makes clear that workers need to better prepare for a future in which AI could play a significant role in many workplaces that until now have been largely untouched by technological disruption.

For people in tech, it means they may be building their AI replacements.

“There’s no question the workers who will be impacted most are those with college degrees, and those are the people who always thought they were safe,” said Burning Glass Institute president Matt Sigelman.

For hundreds of corporations, the researchers estimated the share of payroll spending that goes to workers employed in the 200 occupations most likely to be affected by generative AI. Many of those jobs are held by affluent college graduates, including business analysts, marketing managers, software developers, database administrators, project managers and lawyers.

Companies in finance, including Goldman Sachs, JPMorgan Chase and Morgan Stanley, have some of the highest percentages of their payrolls likely to be disrupted by generative AI. Not far behind are tech giants like Google, Microsoft and Meta.

Getting AI to do human work could result in big savings for those companies. The research estimates that banks and some tech companies spend 60 per cent to 80 per cent of their payrolls, or more, on workers in occupations most likely to be affected by the new technology.

The retail, restaurant and transportation industries are least likely to be affected by generative AI, the report found. Companies like Walmart, McDonald’s and Delta Air Lines mostly employ workers without college degrees who perform roles like helping customers, stocking shelves, cooking food and handling baggage. They spend less than 20 per cent of their payrolls on employees in occupations most likely to be affected by generative AI.

The report doesn’t predict potential job losses related to generative AI. That will be up to employers, the report said, and whether they want to bank the savings from AI automation or use that money to invest and grow, adding more workers. Most experts expect that AI will mostly change jobs for the next few years rather than eliminate them – though that could change if the technology improves sharply.

The report highlights the need for increased training to prepare workers to adapt to a fast-arriving technology, SHRM chief executive Johnny C. Taylor Jr. said.

“Corporations and governments are going to have to seriously invest to get ahead of this,” he said.

The report is the latest entry in a growing field of work trying to predict the effect of generative AI on the economy and the workplace. Other studies have forecast a surge in economic growth and productivity, automating activities that add up to the equivalent of millions of jobs, and time savings of up to 50 per cent for routine office and coding tasks.

In its research, the Burning Glass Institute started with the estimates of generative AI exposure by occupation in a widely cited academic paper that was published last year. It then added its own data sets – including job listings, payroll information, government statistics and corporate disclosures – for the company-by-company calculations. The SHRM report includes a ranking of selected companies. The Burning Glass Institute did the percentage estimates of the payroll spending by company for The New York Times.

Mr Manav Raj, a co-author of the academic paper that the Burning Glass Institute relied on, said the new research appeared to be a credible effort to parse company-level data. But at this stage, he said, all the studies are educated guesses.

“The many papers out there generally conclude that this wave of AI has the potential to have a very large effect,” said Mr Raj, an assistant professor of management at the Wharton School of the University of Pennsylvania. “But it’s going to take some time to find out what that effect really looks like.” NYTIMES