Is artificial intelligence on the verge of surpassing human intelligence? OpenAI CEO Sam Altman believes so. During a panel event at Technische Universität Berlin, Altman suggested that GPT-5 could exhibit intelligence beyond his own. While he expressed optimism about AI’s role in scientific breakthroughs, he avoided specifics on research partnerships and what “smarter than humans” truly means in practice.
With AI models evolving rapidly, the key question remains—will GPT-5 genuinely redefine intelligence, or is this just another ambitious claim?
Altman took some time to engage the audience. He started by asking among them who considered themselves smarter than GPT-4 and who thought they would still surpass GPT-5. The CEO suggested optimism toward AI development which would lead to substantial scientific breakthroughs and innovative research discoveries. His response seems to be somehow scripted, since he avoided answering research partnership questions that were directed to him from the host.
Even while showing his excitement Altman neglected to provide an explanation about what being "smarter" means in practical terms. If intelligence is measured by access to large amounts of data, then GPT-4 already surpasses human knowledge in so many fields. But its limitations in instruction-following, context retention and response revision prevents it from being perceived as truly intelligent.
One crucial aspect in consideration is whether GPT-5 will address these weaknesses. Competing AI models such as DeepSeek have shown their improvements in instruction adherence and response consistency. These qualities contribute to a more efficient AI experience and particularly for all tasks that require structured outputs. If GPT-5 is able to advance in these important capabilities, then it could easily transform how AI is judged in terms of intelligence.
Altman has frequently linked AI to scientific advancements. He has even at some point mentioned the potential it has in curing cancer. However, the connection between language models like ChatGPT and major scientific breakthroughs remains unclear. ChatGPT is officially described as a brainstorming tool, a meeting summarizer and a code assistant. Although these functions are valuable, they do not directly contribute to research tackling unknown scientific questions.
Altman’s vision of AI shaping the future is ambitious, but without concrete details, it remains speculative. While GPT-5 may introduce significant improvements, its practical impact on research and innovation will depend on whether it can truly enhance problem-solving beyond existing AI limitations.