Openai Chief Executive Officer Sam Altman has defended the company’s latest language model, GPT-5, saying that it is a major step in artificial intelligence despite the early wave of criticism. A month after its Rocky Debut, Altman claims that the new model is already helping researchers progress in scientific studies and insists that Openi’s artificial general intelligence (AGI) is very much alive, wired said.

When the GPT-5 was unveiled in August, the launch did not happen as easily as Openai expected. Live Demo was hit by technical issues, and users later took to the reddit to complain that the new model looked less attractive than GPT-4. Many critics accused the company of overseeing their abilities and claimed that the system failed to meet expectations. Some also suggested that AI Boom started cooling. But Altman says the initial response missed what GPT -5 was designed to do. “Vibes were badly bad at the launch,” he said in an interview, “but now they are great.”

GPT 5 focuses on science and research

Altman says that GPT-5 is not only one more chatbot upgrade, but is a model designed to help scientists and researchers in real-world applications. According to him, GPT-5 is beginning to demonstrate capabilities that can accelerate search in physics, biology and mathematics. “GPT-5 is the first time people are, ‘Holy nonsense. It is doing this important piece of physics.” Or a biologist is saying, “Wow, it really helps me understand this,” Altman said. Although Openai has not shared the details of the examples, he believes that it is a significant turn where AI begins to play an active role in advancing human knowledge.

OpenaiI President Greg Brockman said that the impact of GPT -5 may not be clear for regular users as its strength is in particular areas. “Most people are not researchers in physics,” Altman said, “S’s explaining why casual users cannot immediately notice at its depth. Mark Chen, head of Openi’s leading research, reported that the new model now performs at one level compared to the top few Mathematics Olympians, while GPT-4 is given a very low space, indicating its sharp logic skills.

Walking beyond big models

A major argument of critics is that GPT -5 reflects the boundaries of scaling, the notion that the size of the model improves their intelligence automatically. But openi disagree. Ultman and Brockman say that the progress of GPT -5 came not by size but with clever training. The company used reinforcement learning, where human experts guide the model, and even allowed GPT -5 to create their data to refine their learning. “When the model is dumb, what you want to do is trained by a large version of it,” Brockman explained. “When the model is smart, you want to take a sample from it. You want to train on your data.”

This does not mean that the Openai is stepping back from the scale. The company is investing billions in large -scale data centers in Texas and other places to prepare for future versions. Brockman compared the process of rocket science, stating that each upgrade is difficult to construct rapidly, but it is necessary to carry forward the boundaries of what AI can achieve.

Openai wants to redefine the path for Agi

Openai has also begun to reconsider how it talks about AGI – the idea of ​​an AI system that can improve humans in most areas. Instead of considering it as a final goal, the company now calls it a continuous process. “We had a category of error about Openai as a project with a defined end date,” Brockman said. “Now it is about the constant impact and the economy is turning into an AI-managed world.”

Ultman agrees and says that his personal approach to AGI has also developed. Now he believes that real success will come from scientific understanding instead of changing human work only. “Scientific progress definition is actually a big thing for the world,” he said. Altman stated that GPT-5 may not yet be able to “do meaningful science”, but reflects the initial signs of what the future models can achieve.

Altman says GPT 6 will be better

Despite the initial criticism, Altman is confident of the roadmap of Openai. He dismissed the comments of AI researcher Gary Marcus, who claimed that GPT -5 failed to fulfill his promises. “I can tell you with confidence that GPT -6 will be much better than GPT -5, and GPT -7 will be much better than GPT -6,” Altman said, OpeniI’s track record speaks for itself.