A report by OpenAI, the company behind ChatGPT, identified forty-four professions that are exposed to replacement by advanced AI systems, which has raised concerns about the stability and future of employment in various sectors.
The analysis was based on a blind comparison methodology called GDPval, in which human experts evaluated tasks performed by both AI models and professionals, without knowing the authorship of each result.
The evaluators selected the option they considered superiorwhich made it possible to objectively measure the performance of artificial intelligence against human workers in nine key economic sectors in the United States.
Among the models analyzed, Anthropic’s Claude Opus 4.1 managed to outperform human professionals in 47.6% of caseswhile OpenAI’s GPT5-high model achieved 38.8% success.
Among the forty-four professions identified as most vulnerable, some have very high replacement rates. Counter and rental employees turned out to be the most exposed, with a rate of 81% in favor of AI.
Sales managers come secondbeing passed in 79% of the tests. Shipping and receiving managers reach 76%, while editors are surpassed in 75% of cases.
Even professions traditionally considered eminently humanlike private detectives and investigators, show great vulnerability, with a replacement rate of 70%.
The impact of automation is not homogeneous in all sectors. Retail trade tops the list of those most affected, with chatbots that outperform professionals in 56% of the tasks evaluated.
Wholesale trade follows with 53%, and public sector jobs, such as compliance officers and social workers, record a 52% success rate for AI.
In contrast, the information sector, which includes directors, film producers and journalistsshows greater resistance, because even the most advanced model only outperforms humans in 39% of cases.
The OpenAI report provided concrete cases of tasks where AI demonstrated superior performance. Registered nurses were asked to evaluate images of skin lesions and write consultation reportswhile manufacturing engineers were asked to create 3D models of cable reels.
In both cases, the AI systems achieved results that evaluators said equaled or surpassed human work. However, OpenAI warns that the evaluation is limited to specific tasks and that the reality of each profession implies greater complexity than the sum of isolated activities.
The list prepared by OpenAI shows the success rate of AIin front of human professionals in the following occupations:
- Counter employees: 81%
- Sales managers: 79%
- Shipping, receiving and inventory employees: 76%
- Publishers: 75%
- Software developers: 70%
- Detectives and private investigators: 70%
- Responsible for regulatory compliance: 69%
- First-level supervisors of non-retail sales workers: 69%
- Wholesale and manufacturing sales representatives (except technical and scientific products): 68%
- General operations managers: 67%
- Medical and health services managers: 65%
- Buyers and purchasing agents: 64%
- Personal financial advisors: 64%
- Administrative services managers: 62%
- Customer service representatives: 59%
- First-level supervisors of retail sales workers: 59%
- First-level supervisors of production and operations workers: 58%
- Specialist nurses: 56%
- Real estate brokers: 54%
- News analysts, reporters and journalists: 53%
- Computer and information systems managers: 52%
- First-level police and detective supervisors: 49%
- Representatives of wholesale and manufacturing sales of technical and scientific products: 47%
- Lawyers: 46%
- Project management specialists: 42%
- Childhood, family and school social workers: 42%
- Medical secretaries and administrative assistants: 42%
- Stockbrokers, commodities and financial services: 42%
- First level administrative and office support supervisors: 41%
- Financial investment analysts: 41%
- Animators and reenactors: 37%
- Registered nurses: 37%
- Property managers, real estate and community associations: 34%
- Financial managers: 32%
- Producers and directors: 31%
- Audio and video technicians: 30%
- Hotel concierges: 29%
- Order management employees: 28%
- Real estate agents: 27%
- Pharmaceuticals: 26%
- Accountants and auditors: 24%
- Mechanical engineers: 23%
- Industrial engineers: 17%
- Film and video editors: 17%
Despite these projections, OpenAI clarified that the goal of the study is not to claim that humans will imminently lose their jobs.
