Study: GenAI costs billions but doesn’t boost productivity and wages

Generative AI chatbots such as ChatGPT or Gemini have had almost no impact on salaries and jobs thus far, a new study says, asking whether the huge investments in AI models are really worth their while.

We all know the story by now: AI will take jobs from some and depress wages for others. This almost seems impossible to avoid, and we regularly find lists of occupations supposedly facing extinction.

However, that might not be the case at all, at least if we believe the findings of a new study by Danish economists Anders Humlum and Emilie Vestergaard, which examines the labor market impact of AI chatbots on 11 occupations.

All of them have been described as vulnerable to the rise of generative AI – accountants, customer support specialists, financial advisors, HR professionals, IT support specialists, journalists, legal and marketing professionals, office clerks, software developers, and, finally, teachers.

Researchers who covered 25,000 workers and 7,000 workplaces in Denmark in 2023 and 2024 point out that AI chatbot use is now encouraged by most employers who want to boost productivity.

However, according to this working paper, the data simply doesn’t support the hype. That’s because the labor and wage impact of AI chatbots appears to be minimal.

“AI chatbots have had no significant impact on earnings or recorded hours in any occupation. <...> Our findings challenge narratives of imminent labor market transformation due to Generative AI,” the researchers say.

Employees aren’t avoiding the technology, to be clear. On the contrary, their bosses are encouraging them to use the chatbots at work, and 30% of employees have received training, meaning that they’ve adopted the tool – even if they were pushed to do it.

The problem is that the use of chatbots in-house hasn’t yet translated into actual economic benefits.

“Comparing workplaces with high versus low rates of AI chatbot usage, we find no evidence that firms with greater chatbot adoption have fared differently in total employment or wage bills,” says the paper.

While some workers did report substantial benefits from AI chatbots, for instance, saving time, the researchers didn’t detect significantly better labor market outcomes for them.

Moreover, direct survey responses from non-users have confirmed that they have perceived no chatbot-related changes in earnings.

Actually, workers overwhelmingly reported that AI chatbots have not meaningfully affected their earnings: the average perceived earnings impact ranges from 0.04% to 0.2%, depending on whether their employer encourages their use.

This should concern the tech industry that has touted AI as the face and engine of a new industrial revolution. The authors of the study plainly write: “We lack evidence on Generative AI’s economic impacts outside laboratory settings and case studies.”

“My general conclusion is that any story that you want to tell about these tools being very transformative needs to contend with the fact that at least two years after [the introduction of AI chatbots], they've not made a difference for economic outcomes,” Humlum told The Register.