In the interview with Sam Altman by Cleo Abram, the CEO of OpenAI discusses the recent release of Chat GPT-5 and the future of artificial intelligence. The conversation explores the potential impacts of AI on society.
The Power and Capabilities of GPT-5
- Altman states that GPT-4 was the “dumbest model any of us will ever have to use again” and describes GPT-5 as “quite remarkable,” with much-improved capabilities in various tasks, including writing and coding.
- He highlights an example of using GPT-5 to recreate the game of Snake for a TI-83 calculator in seven seconds, demonstrating the model’s ability to create software almost instantaneously.
The Path to Superintelligence and Future Challenges
- Altman predicts that a large language model will make a significant scientific discovery within the next two years. He defines superintelligence as a system that can exceed human capabilities in almost every field.
- He discusses the blurring line between real and fake media, and acknowledges that some jobs will disappear, but believes that young people will have new opportunities for creation and innovation. He expresses more concern for older workers who may be less willing to retrain.
Key Limiting Factors in AI Development Altman identifies four main limiting factors for AI development:
- Compute: The demand for AI far outstrips the current supply of compute power.
- Data: Models are so advanced that they need to be taught how to discover new things and create their own new data.
- Algorithmic Design: Altman believes there are still many “orders of magnitudes of algorithmic gains ahead.”
- Product Integration: Scientific progress is limited if it isn’t integrated into products that people can use.
Reflections and Final Advice
- Altman notes the “vertigo-inducing” rate of change and the need for humility and new solutions.
- He believes that society has a shared responsibility in shaping the future and advises people to prepare for the future by using AI tools to get fluent with their capabilities.