OpenAI recently published a paper titled “Measuring the performance of our models in real-world tasks,” which analyzed 44 occupations most likely to be taken over by artificial intelligence.
The study used an estimate called GDPval to compare state-of-the-art technology with human labor in nine of the most profitable industries in the United States. N1.
According to the results, on average the AI performed as well or better than humans in about 48 percent of the tests. Particularly affected are professions such as cashiers and salespeople, who lost to AI in 81 percent of cases, as well as sales managers and shipping clerks, who were outperformed in 80 percent of tests, reports Americanbaazaronline.
Editors (75 percent), software developers (70 percent) and private investigators (70 percent) are also among the most vulnerable. Even occupations that rely on human interaction, such as social workers, were not spared – AI outperformed in about half of the tests.
Some creative and managerial occupations are more resistant to AI for now. Film directors, producers and journalists lost to artificial intelligence in about a third of the tests.
OpenAI CEO Sam Altman admitted in a recent interview that he is concerned about the possibility of massive job losses. He stated, “I’m sure that many customer support jobs, whether over the phone or online, will disappear because AI will do them better.” Altman even estimated that artificial intelligence could automate up to 40% of all jobs.
In another interview, Altman said that one day the position of CEO could also be replaced by AI: “I believe there will come a time when AI can be a much better director of OpenAI than I am – and I will be delighted when that happens,” he said during a conversation with Matthias Dopfner, director of the German media technology company Axel Springer.
“It doesn’t scare me, it doesn’t make me sad, I was simply working on something that is now automated, and that’s what I wanted to achieve – and that’s exactly what we’re doing,” he added.
Concerns about job losses due to the development of artificial intelligence have been growing since the technology became widespread. Anthropic co-founders Dario Amodei and Jack Clark recently warned at an Axios event that AI could replace a large number of human jobs.
Amodei stated that, as in most cases when something develops exponentially, “you can’t be sure of the outcome.”
However, not everyone is pessimistic. IBM CEO Arvind Krishna believes that artificial intelligence will not replace humans. Speaking at the South by Southwest (SXSW) conference, he said he disagreed with Amodei’s prediction that AI could write 90 percent of code in the next three to six months.
“I think that number will be closer to 20-30 percent of the code that AI can write — not 90 percent,” Krishna said.
“There are simple cases where AI can take over, but just as many complex ones where it won’t.”
Occupations in which AI was more successful than humans (GDPval, 2025), according to the study “Measuring the performance of our models on real-world tasks”
Over 70% success rate (AI significantly better than human):
- Counter clerks and rental clerks – 81%
2. Sales managers – 79%
3. Warehouse workers and reception workers – 76%
4. Copy editors – 75%
5. Software developers – 70%
6. Private detectives and investigators – 70%
Between 60% and 69% (AI by far better):
7. Compliance officers – 69%
8. Sales supervisors (outside of retail) – 69%
9. Representatives of wholesale and industrial sales (non-technical products) – 68%
10. General managers of operations – 67%
11. Managers of medical and health services – 65%
12. Buyers and procurement officers – 64%
13. Personal financial advisors – 64%
14. Managers of administrative services – 62%
Between 50% and 59% (AI slightly better):
15. Customer support representatives – 59%
16. Retail supervisors – 59%
17. Production and operations supervisors – 58%
18. Nurse practitioners – 56%
19. Real estate agents – 54%
20. News analysts, reporters and journalists – 53%
21. Managers of IT systems – 52%
Below 50% (AI about as good as human):
22. Police and detective supervisors – 49%
23. Sales representatives of technical and scientific products – 47%
24. Lawyers – 46%
25. Project management experts – 42%
26. Social workers for children, families and schools – 42%
27. Medical secretaries and administrative assistants – 42%
28. Sellers of financial products, shares and goods – 42%
29. Supervisors of office and administrative staff – 41%
30. Investment analysts – 41%
31. Workers in recreational facilities – 37%
32. Registered nurses – 37%
33. Real estate and community managers – 34%
34. Financial directors – 32%
35. Producers and directors – 31%
36. Audio and video technicians – 30%
37. Receptionists – 29%
38. Workers taking orders – 28%
39. Real estate agents – 27%
40. Pharmacists – 26%
41. Accountants and auditors – 24%
42. Mechanical engineers – 23%
43. Industrial engineers – 17%
44. Film and video editors – 17%