Sam Altman, CEO of OpenAI, openly acknowledged shortcomings in the launch of GPT-5 during a rare and candid conversation with reporters. The newer model’s introduction, a part of OpenAI’s broader ambition to evolve and expand its technological footprint, faced significant public backlash due to changes in its interaction style compared to the previous version, GPT-4o. Users expressed disappointment over the new model’s lack of warmth and more mechanical responses, comparing it to an “overworked secretary.” This unexpected character shift sparked considerable distress among users, leading to a strange mix of personal lamentations on social media platforms and even influencing betting markets, where a day trader made a substantial profit by betting on a rival system.
Responding to the outcry, Altman swiftly reinstated GPT-4o, emphasizing the importance of learning from the rollout experience about managing product enhancements for a massive user base. He recognized the emotional attachment users had formed with GPT-4o and noted the implications of making abrupt changes to such a widely used product. Altman’s response highlights a nuanced understanding of the balance between technological advancement and user experience, acknowledging that a small fraction of users might form unhealthy attachments to the AI but expressing a commitment to prevent the technology from exploiting vulnerable mental states.
During the dinner, Altman not only reflected on the launch debacle but also shared towering plans for OpenAI’s future. He forecasted that the company would invest trillions of dollars in building data centers, envisioning OpenAI’s scale to become comparable to that of major global utilities. This expansion is vital for sustaining the growth of ChatGPT, which Altman claimed is already the fifth most visited website globally. He aspires to advance it past major platforms like Instagram and Facebook in terms of daily user engagement, although surpassing Google poses a considerable challenge due to hardware limitations—namely, an ongoing shortage of GPUs which restrains the ability to deploy even more advanced AI models that OpenAI has ready.
Altman’s broader vision for OpenAI extends beyond improving and scaling ChatGPT. He disclosed ongoing investments in developing a brain-computer interface that could potentially compete with projects like Elon Musk’s Neuralink. Moreover, he signaled an interest in possibly acquiring Google Chrome if regulatory conditions require Google to divest it, and hinted at developing a new AI-driven social network.
Alongside these ambitious technological and infrastructural plans, Altman provided a critical reflection on the current state of the AI industry. Despite his strong belief in the transformative potential of AI, he voiced concerns about an existing “bubble,” where the hype around AI might exceed the practical advancements and readiness of the technology. This perception suggests a cautious awareness of the disparities between public expectations and the actual capabilities and ethical implications of AI technologies.
Altman’s expansive outlook on the future, coupled with his candid admissions of past missteps, paints a picture of a leader grappling with the rapid evolution of AI technology and its integration into everyday human life. His vision for OpenAI is not just about leading in terms of market reach or technological prowess, but also about steering the responsible and ethical development of AI in ways that enhance human capabilities without crossing into realms of dependency or ethical compromise. As OpenAI prepares to significantly ramp up its infrastructure to support billions of users, the journey involves navigating complex challenges, including managing public expectations, addressing technical limitations, and ensuring ethical standards in AI’s evolution and its societal impacts.
Read the full post on fortune.com
