In this revealing video, Matthew Berman discusses the latest information regarding OpenAI’s secretive project, Q-Star, and its new code name, Strawberry. Drawing from a Reuters article, Berman outlines how Strawberry aims to enhance the reasoning and planning capabilities of AI models, addressing the limitations of current large language models (LLMs) that struggle with complex logical tasks. The video highlights OpenAI’s ambition to develop a system that not only generates answers but also autonomously navigates the internet to conduct deep research. Berman explains the significance of continuous learning and memory in AI, emphasizing the need for models to revisit and reflect on information rather than merely processing it once. He discusses the potential for Strawberry to improve multi-step problem-solving and planning, which are crucial for achieving artificial general intelligence (AGI). The video also touches on OpenAI’s internal framework for assessing the progress of its models towards AGI, outlining five levels of capability, with current models only reaching level one. Berman concludes by speculating on the future of AI and the implications of these advancements for humanity, inviting viewers to share their thoughts on the developments surrounding Q-Star and Strawberry.

Matthew Berman
Not Applicable
August 4, 2024
OpenAI Working on New Reasoning Technology
PT14M16S