OpenAI developing AI project 'Strawberry' to enhance reasoning abilities in AI

From Business Today: 2024-07-13 09:08:55

OpenAI is developing an AI project called “Strawberry” to enhance reasoning abilities in AI, aiming for breakthroughs in autonomous research and problem-solving. The project involves a unique training and processing approach, unlike existing systems, focusing on equipping AI to conduct deep research and multi-step problem-solving autonomously, a significant advancement in the industry.

The goal is to achieve human-level AI reasoning, with major tech companies like Google, Meta, and Microsoft also exploring similar techniques to unlock AI’s potential in driving scientific discoveries and complex software development. OpenAI has not confirmed details about Strawberry but emphasized the continuous research into new AI capabilities to improve reasoning over time, aligning with industry practices.

Strawberry builds upon OpenAI’s Q* project, known for advanced reasoning abilities, with internal reports of solving complex math and science problems beyond current AI capabilities. The project involves a unique form of post-training, refining AI models after training on massive datasets using techniques like fine-tuning and self-generated training data to enhance reasoning abilities effectively.



Read more at Business Today: OpenAI’s secret weapon: New AI reasoning technology under code name ‘Strawberry’