Sam Altman, CEO of OpenAI, stated in February 2026 that his company had "basically built AGI, or we are very close to it". The claim came during an interview with Forbes magazine, in which Altman discussed his journey through the world of artificial intelligence. A few days later, however, he softened his statement. "I meant it as a spiritual statement, not a literal one," Altman said. According to him, achieving AGI requires "a lot of medium-sized breakthroughs", but not one major leap forward.
Altman's words sparked a heated debate in the technology community. David Pierce of The Verge noted that Altman's statement was typically ambiguous—he first declared victory, then backed away. Microsoft CEO Satya Nadella responded skeptically to Altman's claim: "I don't think we're anywhere close to AGI," he said with a smile. Nadella emphasized that deciding whether AGI has been achieved is not up to Altman or himself—it is a process that requires a clear benchmark.
What Does AGI Actually Mean?
Artificial general intelligence (AGI) is a system capable of performing nearly all cognitive tasks that a human can. However, the definition remains ambiguous and inconsistent. According to a March 2025 survey conducted by the Association for the Advancement of Artificial Intelligence, 76% of leading researchers believe that current approaches to AI are unlikely or very unlikely to lead to AGI.
The scientific journal Nature published an article arguing the opposite. A group of scientists from the fields of philosophy, machine learning, linguistics, and cognitive science concluded that current large language models (LLMs) already exhibit general intelligence. According to their analysis, OpenAI's GPT-4.5 passed the Turing test in March 2025, with people identifying it as human in 73% of cases—more often than actual humans.
Evidence of Human-Level Intelligence
LLMs have achieved remarkable results. They have won a gold medal at the International Mathematical Olympiad, collaborated with leading mathematicians on proving theorems, and generated scientific hypotheses that were subsequently verified through experiments. They can solve doctoral exam problems, help professional programmers write code, and communicate with hundreds of millions of people around the world.
According to the Nature article, current LLMs meet two levels of evidence for general intelligence. The first is the Turing-test level—passing basic school exams, conducting reasonable conversations, and performing simple reasoning. The second is the expert level—winning gold medals in international competitions, solving doctoral exam problems across disciplines, writing and debugging complex code, being fluent in dozens of languages, and providing useful assistance with research at the frontiers of knowledge.
Criticism and Objections
Critics argue that LLMs are merely "stochastic parrots" that simply rearrange patterns from training data. Another objection is that AI lacks world models, embodiment, or autonomy. However, the Nature scientists refute these arguments. They point out that current LLMs can solve new, unpublished mathematical problems and demonstrate cross-domain knowledge transfer—for example, training on code improves general reasoning even in areas unrelated to programming.
While some scientists claim that AGI is already a reality, others remain skeptical. The problem is partly conceptual—definitions of AGI are ambiguous. It is partly emotional—AGI raises fears of humans being replaced. And it is partly practical—the term is tied to business interests that may distort assessments. Altman himself acknowledges that his motivations may be confusing to some. His longtime mentor Paul Graham says it is "hard to know what's going on inside his head." Altman's commitment to spend $1.4 trillion (approximately CZK 32 trillion) on AI chips and data centers over the next eight years raises questions about the financial feasibility of these plans.
The debate over whether we have achieved AGI continues. One thing, however, is certain—the boundary between human and artificial intelligence is rapidly blurring.
