‘Future of work’: How OpenAI’s self-improving AI model is preparing to replace research jobs

OpenAI is preparing for a future where artificial intelligence systems could improve themselves without direct human involvement, according to recent reports. The company is now hiring safety researchers to study the risks linked to this development, known as recursive self-improvement, where AI systems train better versions of themselves. This shift comes as AI capabilities continue to grow rapidly, with companies exploring how far automation can go in research and software development.

What is self-improving AI and why it matters

Recursive self-improvement refers to AI systems that can upgrade their own performance without relying heavily on human input. In simple terms, an AI system could build a smarter version of itself, which could then repeat the process. This idea, once limited to theory, is now being actively explored as AI tools improve. Reports suggest that the amount of work AI systems can handle is increasing quickly, with some studies indicating that their capabilities are doubling roughly every seven months. If this trend continues, AI systems could eventually handle complex tasks that currently require human researchers and engineers.

Why OpenAI is hiring safety researchers

OpenAI’s hiring move shows that the company is not only building advanced systems but also preparing for the risks they may bring. The role focuses on studying problems that may arise in the future, even if they do not exist today.

The researchers will work on areas such as:
Preventing data poisoning attacks
Understanding how AI systems make decisions
Studying risks linked to highly autonomous systems
Tracking how AI tools are replacing human technical work
The company has described this work as urgent, with potential impact on both industry and society.

AI automation and impact on jobs

One of the key points in the job listing is the need to track the “automation of technical staff”. This includes measuring how AI tools, such as coding assistants, are replacing or supporting human engineers.

Industry research suggests that AI could soon handle a large share of software tasks that currently take humans days or weeks to complete. This raises questions about the future of jobs in research, engineering and development. At the same time, companies continue to hire experts to manage and guide these systems, showing a shift in the nature of work rather than a complete replacement.

OpenAI’s long-term AI vision

OpenAI CEO Sam Altman has previously outlined the company’s goals, including building an “automated AI research intern” in the near term and a fully automated AI researcher by 2028. While he has acknowledged that these goals may not be achieved, the company continues to invest in this direction, reflecting strong belief in the long-term potential of AI.

Global race in self-improving AI

Other companies are also exploring similar ideas. Anthropic has published research on using AI systems to monitor stronger AI models, while experts estimate there is a 60 per cent chance that AI research could happen without direct human involvement by the end of this decade. This shows that the race towards advanced AI is not limited to one company but is a broader industry trend. The development highlights a key shift in artificial intelligence: moving from tools that assist humans to systems that may operate independently. While this could improve efficiency and innovation, it also raises concerns about control, safety and employment. For now, companies like OpenAI are trying to balance progress with preparation, ensuring that as AI becomes more capable, its risks are understood and managed.