OpenAI is moving faster than ever – and it’s setting its sights on something big.
The company says its deep learning systems are now advancing so quickly that it expects to develop an “AI research assistant” capable of working at an intern level by 2026, and a fully autonomous AI researcher by 2028.
That’s right – an AI that can independently plan, reason, and carry out research projects from start to finish.
A New Era for OpenAI
This bold prediction came during a livestream on Tuesday, where CEO Sam Altman and Chief Scientist Jakub Pachocki also announced a major shift in OpenAI’s structure.
The company is now officially a public benefit corporation (PBC) – moving away from its non-profit roots. This change gives OpenAI more freedom to raise capital and scale its operations while still keeping a public-interest mission at its core.
Altman described the move as a necessary step for OpenAI’s ambitious roadmap, which includes creating AI tools that can accelerate scientific discovery and innovation across industries.
“We’re building the infrastructure to support the next generation of AI research,” Altman said, adding that OpenAI’s new structure balances rapid growth with responsible development.
Building Toward an AI Scientist
So what does it actually mean to have an “AI researcher”?
According to Pachocki, it’s not about building a human-like scientist.
Instead, it’s about creating a system capable of autonomously delivering on complex research projects – something like a highly advanced assistant that can form hypotheses, test ideas, and refine its understanding over time.
He also noted that OpenAI’s models are already showing surprising progress.
Current systems can handle tasks with time horizons up to five hours and perform on par with top human competitors in challenges like the International Mathematical Olympiad.
But OpenAI believes that’s just the beginning.
Pushing the Boundaries of Compute
One of the company’s biggest focuses right now is “test time compute” – essentially, how much time and computing power a model uses to think through a problem.
By extending this “thinking time,” OpenAI expects to push its AI systems toward deeper reasoning and problem-solving abilities.
In fact, Pachocki said the company may eventually dedicate entire data centers to solving just one scientific challenge.
Imagine an AI system spending days or weeks processing a single experiment – that’s the kind of scale OpenAI believes will lead to superintelligent systems within the next decade.
A Massive Bet on Infrastructure
To make this vision real, OpenAI is making a massive infrastructure investment.
Altman revealed that the company has committed to building 30 gigawatts of compute capacity – an estimated $1.4 trillion investment – to power future AI development.
That’s not just a technical goal; it’s a sign of how far OpenAI plans to go to stay ahead in the race for AI dominance.
Keeping Science and Ethics Connected
Even as OpenAI moves into for-profit territory, it’s keeping one foot grounded in research ethics.
The OpenAI Foundation, the company’s non-profit arm, will retain 26% ownership of the for-profit entity. It will oversee research priorities and manage initiatives aimed at using AI for good.
That includes a $25 billion commitment toward using AI to cure diseases and advance global scientific research safely.
“Our goal is to align progress with purpose,” Altman said, reinforcing the company’s promise to develop AI responsibly even as it scales rapidly.
The Race Toward Superintelligence
Pachocki added that OpenAI believes superintelligence – AI systems that outperform humans across many critical areas – could emerge in less than a decade.
While that idea might sound like science fiction, the company insists it’s already laying the groundwork.
Its focus on algorithmic innovation, massive compute power, and safety oversight reflects a careful balance between speed and responsibility.
Still, the announcement raises questions:
- What happens when an AI system can do advanced research faster than any human?
- How will that affect scientific jobs – or the pace of discovery itself?
Those are the debates OpenAI seems ready to lead as it continues to redefine what’s possible with artificial intelligence.
Key Takeaways
| Area | Details |
|---|---|
| Goal | Develop an AI research assistant by 2026 and a full AI researcher by 2028 |
| Company Structure | Now a Public Benefit Corporation (PBC) |
| Investment | $1.4 trillion commitment for 30 gigawatts of infrastructure |
| AI Foundation Role | 26% ownership; oversees ethics and research |
| Focus Areas | Deep learning, test time compute, and responsible AI development |
