Sam Altman thinks AI will have ‘novel insights’ next year

In a new essay published on Tuesday “Delicate loneliness,“Openai CEO Sam Altman shared the latest vision in the next 15 years to see how the EU will change the human experience.

Esse is a classic example of Altman’s futurism: AGI prepare his promise and claimed his visit to the feat, claiming that his company was very close. Often Openai CEO This publishes the essays of natureAGI’s modern work, put a future in violation of the concept of energy and social contract. But often, in general, there are tips on the essays of the Alti Openai on the next.

At a point of construction, Altman claimed that next year, the world will see the future in 2026 [AI] systems that can understand the concepts of novels. “Although this is somewhat uncertain, Openai leaders recently said that the company’s AI models are aimed at prepare new, interesting ideas.

While announcing Openai’s O3 and O4-Mini EU Thinking Models In April, co-founder and President Greg Brockman said that the scientists have the first models used to create new, useful ideas.

Altman’s blog posts, Openai, Openai himself, can increase the efforts to develop the EU, which can create novel concepts. Openai, of course, will not be the only company aimed at this site – several Openai’s rivals, scientists have been focused on new assumptions, and thus new discoveries about the world are aimed at AI models.

In May, Google left a Paper in AlphaorveThe company is an AI coding agent claiming novel approaches to complex mathematics. Another start claiming that the former Google CEO Eric Schmidt, the future of the AI ​​Agent The ability to discover a real scientist. Launched an an anthropic program in May Support scientific research.

If successful, these companies can automate the basic part of the scientific process and enter their nuclei to mass areas and other areas such as drug discovery, material science and science.

This will not be the first time with Altman, Openai’s plans on a blog plans. He wrote Altman in January Another blog post It offers to be an agent year of 2025. After his company, the first three AI agents continued: Operator, Deep researchand The code.

However, the AI ​​systems can be more difficult than making the agent to get to create novel concepts. The wider scientific community approaches the ability to create a true original idea of ​​the AI.

At the beginning of this year, hugging the chief science officer, Thomas Wolf wrote an essay that argues this Modern AI systems cannot ask big questionswhich is the key to any major scientific progress. The former Openai Research Rig was in the TechCrunch, which could not create novel assumptions of today’s AI models in TechCrunch.

Stanley is now building a team of $ 200 million to create a strong laboratory of AI, which is aim to buy EU models specifically for better assumptions of AI to face better assumptions. This is a difficult problem, because it is a sense of making AI models creatively and interesting in Stanley.

Openai really stays in the AI ​​model that can produce a new concept and remains to appear. Again, Altman’s essay can do something familiar – a preview of Openai’s side.