In a new essay published on Tuesday called “The Gentle Singularity”, the Operai CEO, Sam Altman, shared his last vision on how AI will change the human experience in the next 15 years.
The essay is a classic example of Altman’s futurism: promising AGI’s average, and arguing that the company is quite close to the feat, while simultaneusia minimizes its arrival. The OpenAI CEO often publishes essays of this nature, cleanly establishing a future in which AGI interrupts our modern conception of work, energy and social contract. But often, Altman’s essays contain suggestions on what OpenI is working below.
At a time of the essay, Altman said that next year, in 2026, the world “will probably see the arrival of [AI] Systems that can find new ideas. While this is vague, Openi executives have recently indicated that the company focuses on making the models of new and interesting ideas about the world.
When announcing the O3 and O4-Mini Ai AI models in April, the co-founder and President Greg Brockman said that these were the first models that scientists had used to generate new and useful ideas.
Altman’s publication suggests that in the next year, Openai Itelf can increase its efforts to develop an AI that can generate new ideas. Openai would certainly be the only company focused on this effort: several of Openai competitors have changed their approach to train AI models that can help scientists come with new hypotheses and, therefore, new discoveries over the world.
In May, Google launched an article about Alphaevevve, an AI coding agent that the company claims to have generated novel approaches for complex mathematical problems. Another startup backed by former Google CEO, Eric Schmidt, Futurehouse, states that his AI agent tool has been capable or making a genuine scientific discovery. In May, Anthrope launched a program to support scientific research.
If you succeed, these companies could convey a key part of the scientific process and enhanced in mass industries such as the discovery of drugs, material science and other fields with science in their nucleus.
This would not be the first time that Altman gave him a hat on OpenI’s plans in a blog. In January, Altman wrote another blog publication suggestion that 2025 would be the year of agents. Then, his company prosecuted to leave his first three AI agents: operator, deep research and Codex.
But making AI systems generate novel ideas may be more difficult than making them agents. The broader scientific community remains skeptical of AI’s ability to generate genuinely original ideas.
Earlier this year, Hugging Face science director, Thomas Wolf, wrote an essay arguing that modern AI systems cannot ask great questions, which is key to any great scientific breakup. Kenneth Stanley, former protagonist of Operai research, also previously told TechCrunch that today’s models cannot generate novel hypotheses.
Stanley is now building a team in Lilac Sciences, a startup that raised $ 200 million to create a laboratory specifically focused on making the models of AI come with better hypotheses. This is a difficult problem, according to Stanley, because it implies giving the models of AI a sense of what is creative and interesting.
If Openai really creates an AI model that is capable or that produces novel ideas is still soe. Even so, Altman’s essay can present something familiar: a preview of where OpenAi is directed below.
