Jakub Pachocki’s Vision: How OpenAI’s New Chief Scientist Sees AI’s Future

From Theoretical Computer Science to Leading AI’s Next Frontier

Jakub Pachocki, OpenAI’s newly appointed chief scientist in 2024, has spent years bridging the gap between abstract theory and practical AI breakthroughs. After joining the company in 2017 from academia—where he specialized in theoretical computer science—Pachocki now helms research at the firm behind ChatGPT, the viral chatbot launched in 2022. But his ambitions stretch far beyond conversational AI. OpenAI’s toolkit now includes reasoning models that tackle logic-heavy tasks, assist researchers in coding, and even generate scientific hypotheses. Yet as the company scales, it faces mounting scrutiny over the energy appetite of its models and its reliance on proprietary systems that limit external innovation.

“AI is evolving from an assistant to an autonomous researcher,” Pachocki says. “Our Deep Research tool can already operate unsupervised for 10–20 minutes, though today’s compute footprint is modest.”

Pachocki envisions AI soon mastering autonomous research in fields like software engineering and hardware design. The key, he argues, lies in merging pre-training with reinforcement learning—a combo that could let models iteratively refine their outputs without constant human oversight. But he’s quick to clarify that AI cognition diverges sharply from human thought: these systems lack any awareness of how or when they acquired knowledge, operating as “black boxes” with unparalleled pattern-matching prowess.

OpenAI’s proprietary approach has drawn fire from researchers who argue that closed models stifle collective progress. Pachocki acknowledges the tension, noting plans to release an open-weight model—a rare concession in a portfolio dominated by locked systems. For now, the company’s focus remains on pushing boundaries, even as debates over ethics, accessibility, and sustainability loom larger than ever.