How OpenAI is betting on one powerful “brain” over many tools

This article summarizes an episode of Alex Kantrowitz’s video series featuring Greg Brockman, president and co-founder of OpenAI.

Greg Brockman, president and co-founder of OpenAI/ Photo credit: Alex Kantrowitz

Building automated systems that can match human reasoning requires massive amounts of computing power. Greg Brockman, president and co-founder of OpenAI, argues that developing multiple technologies divides these limited resources.

He warns that pursuing every promising idea slows down progress toward creating artificial general intelligence (AGI).

To maintain momentum, Brockman believes engineering teams must prioritize core reasoning models over specialized applications. By focusing the available hardware on text-based systems, the company ensures its main projects receive enough support to solve difficult real-world problems.

Dealing with limited resources

Allocating hardware to different research divisions forces companies to abandon certain projects. The current landscape offers too many opportunities, as many ideas yield practical results. This can eventually strain available computing power, forcing difficult leadership decisions.

To manage these constraints, the company halted development on its video generation software.

Brockman explains this shift, “They are built in a different way, and to some extent, we are saying that pursuing both branches is hard for us to do for these applications.”

Focusing on language systems

Directing all funding toward text-based programs requires certainty that these models can solve complex problems. Past debates over the limits of text intelligence are now considered settled internally.

Upcoming models are expected to prove these systems can develop a true understanding of how the world operates.

Real-world trials reinforce this approach. Brockman shared, “A physicist had been working on a problem for some time… Twelve hours later, we have a solution,” calling it the first time it felt like the model was truly “thinking.”

Designing the user experience

Translating these text models into daily tools requires building an interface that connects the technology to user routines.

The company focuses on three distinct steps to achieve this:

• Combining different tools into a single platform.
• Building systems that remember past user instructions.
• Creating an interface that links the technology to standard computer operations.

Upgrading internal systems

Establishing a timeline for AGI