🚀 The New Era of Computing: Sam Altman on GPT-5, Superintelligence, and the Challenges Ahead

GPT-5

Sam Altman, CEO of OpenAI, recently spoke in an interview following the announcement of GPT-5, discussing the next steps in artificial intelligence, the societal changes we can expect, and the challenges on the path towards superintelligence.

Altman views the evolution of AI as a new era where GPT-4 was the dumbest model we will ever have to use.

The Enhanced Capabilities of GPT-5

GPT-5 represents a significant step up from its predecessor. Altman highlights the model's exceptional performance in the following areas:

• Scientific and Technical Queries: The model provides much more accurate and in-depth answers to complex, technical problems.

• Instant Software Development: In terms of coding tasks, GPT-5 is most impressive, capable of generating on-demand software almost instantaneously.

• More Natural Communication: The model handles linguistic nuances much better, offering a smoother and more "natural" user experience.

On the Brink of Superintelligence

Altman shared his vision for what comes next in the journey toward superintelligence. For an AI to make a real scientific breakthrough, it must progress from solving "tasks that take a few minutes" to performing thousands of hours of research work.

He believes superintelligence will be the AI system that could perform research and run OpenAI better than humans. Significant scientific discoveries could occur within the next couple of years if the pace of development is maintained.

A Glimpse into 2030

The interview included a hypothetical time travel exercise, in which Altman outlined how the world might change in the coming years:

Area	Expected Change by 2030
Media and Reality	Due to artificially generated media (deepfakes, synthetic content), distinguishing reality will become difficult, but people will become more tolerant of what is considered "real enough."
Job Market	Although some jobs will disappear, AI will create unprecedented opportunities for young people to be creative and found new, high-value companies.
Healthcare	GPT-5 is already excellent in providing health advice. Altman hopes that by 2035, AI will be able to cure or treat a significant number of diseases.

The Limits to Progress: Not Data, But Energy

Altman states that four main constraints currently impede AI development: Compute, data, algorithmic design, and product building.

However, he identifies the biggest challenge not as data or algorithms, but as the physical infrastructure. Ensuring the energy required to power gigawatt-scale data centers is the most significant obstacle.

The Ultimate Advice

The tech leader advises everyone to use the tools and experiment with them. This is the best way to prepare for change and leverage the new, vast opportunities offered by AI.