AI to Unlock New Discoveries in 2026, Says OpenAI CEO

In a recent essay titled "The Gentle Singularity," published on Tuesday, OpenAI CEO Sam Altman outlined his vision for AI's transformative impact on humanity over the next 15 years.

Altman's essay blends optimism about artificial general intelligence (AGI) with a tempered view of its arrival. Known for such forward-looking writings, he envisions AGI reshaping work, energy, and societal structures. His essays often offer glimpses into OpenAI's upcoming projects.

In the essay, Altman predicted that by 2026, AI systems will likely deliver novel insights. Though not specific, this aligns with OpenAI's recent focus on developing AI models that generate innovative ideas about the world.

In April, OpenAI's co-founder and President Greg Brockman announced the o3 and o4-mini AI reasoning models, noting that scientists had used them to produce valuable new ideas.

Altman's post hints that OpenAI may intensify efforts to create AI capable of generating novel insights in the coming year. Other companies, including Google with its AlphaEvolve AI coding agent and FutureHouse, backed by former Google CEO Eric Schmidt, are also pursuing AI-driven scientific discoveries. In May, Anthropic launched a program to advance scientific research.

Success in this area could revolutionize fields like drug discovery and material science by automating critical parts of the scientific process.

Save Over $200 on Your TechCrunch All Stage Pass

Innovate smarter. Grow faster. Network deeper. Join leaders from Precursor Ventures, NEA, Index Ventures, Underscore VC, and more for a day of strategies, workshops, and impactful connections.

Save Over $200 on Your TechCrunch All Stage Pass

Innovate smarter. Grow faster. Network deeper. Join leaders from Precursor Ventures, NEA, Index Ventures, Underscore VC, and more for a day of strategies, workshops, and impactful connections.

This isn’t the first time Altman has hinted at OpenAI’s plans in a blog. In January, he suggested 2025 would be the year of AI agents, and OpenAI later released three agents: Operator, Deep Research, and Codex.

However, creating AI that generates novel insights may be tougher than building agentic systems. Some in the scientific community question AI’s ability to produce truly original ideas.

Earlier this year, Hugging Face’s Chief Science Officer Thomas Wolf argued that current AI systems struggle to ask profound questions, a cornerstone of scientific breakthroughs. Kenneth Stanley, a former OpenAI research lead, also told TechCrunch that today’s AI models cannot generate novel hypotheses.

Stanley now leads a team at Lila Sciences, a startup that raised $200 million to develop an AI-powered lab focused on improving AI’s ability to form hypotheses. He notes that this challenge requires instilling a sense of creativity and curiosity in AI models.

Whether OpenAI can deliver an AI model capable of producing novel insights remains uncertain. Yet, Altman’s essay likely offers a preview of OpenAI’s next steps.