OpenAI CEO Sam Altman has made a striking claim about the rapid development of artificial intelligence, saying that advanced AI systems have progressed to a point where they are now outperforming some of the world’s most capable scientists in many areas.
Speaking on Monday at the opening ceremony of the Abu Dhabi International Petroleum Exhibition and Conference (ADIPEC) in the United Arab Emirates, Altman described how dramatically AI technology has evolved over the past year. He suggested that the progress has been so significant that tasks once considered beyond the capabilities of artificial intelligence are now becoming increasingly manageable.
During his address in Abu Dhabi, Altman compared the capabilities of AI systems at the beginning of the year with their current performance.
According to the OpenAI chief executive, the difference has been substantial, particularly when it comes to scientific reasoning and solving complex problems.
“Earlier this year, artificial intelligence was mostly not as smart as our smartest scientist, and now it’s mostly smarter than them,” Altman said.
Scientific research is one area where this development could have major implications. AI tools can help researchers examine large datasets, identify patterns, develop hypotheses and explore possible solutions to problems that might otherwise take significant amounts of time.
However, outperforming scientists in certain tasks does not necessarily mean that AI can independently replace scientific expertise. Research still depends on experimental verification, careful interpretation of evidence and the ability to determine whether a seemingly convincing result is actually correct.
AI Is Taking on More Professional and Office-Based Tasks
Altman also discussed how artificial intelligence has moved beyond the relatively limited capabilities it demonstrated in earlier stages of development.
In the past, AI systems struggled with many tasks associated with professional occupations, particularly those requiring several steps, complex reasoning or the ability to work with different types of information.
That situation is changing rapidly.
Altman suggested that modern AI models can now perform a substantial portion of the work typically associated with office-based professions.
These responsibilities can include preparing documents, analyzing information, organizing research, assisting with software development and handling various administrative assignments.
For businesses, the potential benefits are significant. Companies may be able to complete certain projects more quickly, reduce repetitive workloads and give employees more time to concentrate on responsibilities that require direct human involvement.
Nevertheless, the ability to perform individual tasks is different from being able to take responsibility for an entire job. Most professional positions involve communication, decision-making, accountability and an understanding of circumstances that cannot always be reduced to a series of instructions.
As a result, the expansion of AI into the workplace is likely to involve a combination of automated processes and human supervision rather than a straightforward replacement of employees.
Why Altman Believes Humans Must Remain in Control
Despite his optimistic assessment of AI’s capabilities, Altman emphasized that technological progress should not come at the expense of human decision-making.
He argued that people must retain control over the direction in which artificial intelligence develops and how it is used in society.
“We think it’s absolutely important that humans remain in control of the future; there shouldn’t be a shift to AI, there shouldn’t be a ‘we’ll let AI decide what’s best for us,'” he stated.
As these technologies become more capable, questions about oversight, responsibility and the limits of automated decision-making are becoming increasingly important.
For example, AI can provide recommendations in areas such as healthcare, finance, education and business management. However, decisions in these fields can have serious consequences, making human judgment and accountability essential.
Altman reinforced this position by describing artificial intelligence as a technology that should serve people rather than operate independently of their interests.
“AI should be a tool built for people and should serve them, not the other way around,” Altman added.
His position suggests that the future of AI should not simply be measured by how much work machines can perform, but also by whether their capabilities improve people’s lives while preserving meaningful human control.
Energy Supply Could Become the Biggest Obstacle to AI Development
Beyond the technological and social implications, Altman also highlighted a more practical challenge facing the artificial intelligence industry: the enormous amount of electricity required to operate modern AI infrastructure.
The growing popularity of AI tools has increased demand for computing capacity. Behind every advanced AI model is an extensive network of data centers, computer processors, cooling systems and other infrastructure that requires substantial amounts of energy.
Training powerful models and responding to millions of user requests also place significant demands on computing resources.
Altman identified energy availability as one of the most important limitations that could determine how quickly the industry expands in the coming years.
He suggested that while technological improvements are likely to continue across many areas, the physical resources needed to support that progress cannot be developed at the same pace without addressing energy requirements.
“We will make great progress in everything else, except that atoms are still atoms and the physical world that needs to power this will be the dominant factor of cost and abundance,” he concluded.
