TLDR: OpenAI CEO Sam Altman has robustly defended GPT-5, launched in August 2025, against initial criticisms regarding its performance and unmet expectations. He emphasizes its significant advancements in scientific AI, accelerating discoveries in fields like physics and biology, and clarifies that OpenAI’s vision for Artificial General Intelligence (AGI) is evolving towards AI-driven scientific understanding rather than merely replacing human labor. While acknowledging early ‘bad vibes,’ Altman asserts GPT-5 is a crucial step, though not yet full AGI due to its inability to continuously learn post-deployment.
OpenAI’s latest language model, GPT-5, launched in August 2025, has been met with a mix of strong debate and reactions within the AI community. Despite an initial wave of criticism and technical setbacks during its debut, OpenAI CEO Sam Altman has come forward to staunchly defend the model, characterizing it as a major leap forward, particularly for scientific artificial intelligence and the ongoing pursuit of Artificial General Intelligence (AGI).
Altman asserts that GPT-5 is not merely another chatbot upgrade but a sophisticated tool designed to assist scientists and researchers in real-world applications. He highlighted that the model is already demonstrating capabilities that could accelerate discovery in fields such as physics, biology, and mathematics. “GPT-5 is the first time where people are, ‘Holy fuck. It’s doing this important piece of physics,'” Altman reportedly stated, indicating a turning point where AI meaningfully contributes to scientific breakthroughs. OpenAI’s head of research, Mark Chen, further supported this by noting GPT-5’s performance is comparable to top Math Olympians, a significant improvement over its predecessor, GPT-4.
The initial rollout of GPT-5 was not without its challenges. Altman admitted that “The vibes were kind of bad at launch,” citing technical glitches during OpenAI’s live demo and user complaints about the model feeling less engaging than GPT-4. Critics, including AI researcher Gary Marcus, accused OpenAI of overpromising and failing to deliver on expectations, with some even speculating about a fading excitement around generative AI. However, Altman believes these judgments overlooked the model’s core purpose, which lies in specialized optimizations for tasks like coding and science, benefiting experts more than everyday users. OpenAI clarified that GPT-5’s advancements stem from reinforcement learning and training on its own data, rather than just scaling up datasets and computation.
Furthermore, Altman elaborated on OpenAI’s evolving perspective on AGI. He explained that the company’s view of AGI has shifted from a fixed endpoint to a continuous process focused on transforming the global economy through AI-powered scientific understanding, rather than simply replacing human work. “The scientific progress definition is really a big deal for the world,” he emphasized, suggesting that true breakthroughs will emerge from AI-driven discovery.
Also Read:
- OpenAI Poised to Unveil Advanced Agent Builder at Upcoming DevDay Event
- Microsoft Introduces GPT-5 Agent Mode to Enhance Productivity Across 365 Applications
Despite his strong defense of GPT-5’s capabilities, Altman also clarified that the model does not yet meet the full criteria for AGI. He acknowledged that while GPT-5 exhibits a high level of general intelligence, it still lacks crucial components, particularly its inability to continuously learn independently from new experiences once deployed. “This model shows a level of general intelligence, yet we still seem to be missing a few essential components that are crucial to AGI,” Altman stated. Nevertheless, he remains confident about OpenAI’s future roadmap, asserting, “What I can tell you with confidence is GPT-6 will be significantly better than GPT-5, and GPT-7 will be significantly better than GPT-6.”
