OpenAI’s ChatGPT Upgrade, Codenamed Strawberry, Aims to Enhance Reasoning Abilities

OpenAI, the leading artificial intelligence research organization based in San Francisco, is reportedly working on a significant upgrade for its popular chatbot, ChatGPT. Codenamed “Strawberry,” this project aims to enhance the chatbot’s reasoning abilities and enable it to perform deep research operations by autonomously navigating the internet.

According to a leaked internal document dating back to at least May, Strawberry is being developed secretly within OpenAI. The document suggests that the Strawberry models will allow ChatGPT to plan ahead and conduct complex internet searches reliably, catering to users’ needs for in-depth research.

Strawberry, which appears to be an evolution of a previous internal codename called Q*, has been described as a major breakthrough in AI. Sources claim that the upgraded model, possibly Strawberry, demonstrated the ability to answer challenging science and math questions that are currently beyond the reach of existing chatbots.

While the exact details of Strawberry remain undisclosed, recent reports indicate that OpenAI showcased a project with “human-like reasoning” during an internal all-hands meeting. It is uncertain whether this project aligns with the Strawberry upgrade.

OpenAI has also introduced a new classification system for AI models, ranging from Level 1 to Level 5. ChatGPT currently represents Level 1, while Level 5 signifies AI capable of performing the work of an entire organization. Notably, Level 3 involves AI that can take action on a user’s behalf over several days, while Level 4 denotes AI capable of independent innovation.

The Strawberry project, as mentioned in the leaked document, aims to equip ChatGPT with the ability to perform long-horizon tasks (LHT). These tasks require the model to plan ahead and execute a series of actions over an extended period, utilizing computer-using agents (CUA) to autonomously research the web and act based on the findings.

OpenAI intends to test these advanced models on the work of software and machine learning engineers, potentially leading to AI training and the development of future AI variants on the path to achieving artificial general intelligence (AGI).

While the exact release date for Strawberry remains uncertain, OpenAI has reportedly informed developers and other stakeholders that it is on the verge of deploying AI with improved reasoning abilities. Although it is unclear if ChatGPT will receive upgrades this year, the rapidly evolving AI industry compels OpenAI to introduce new innovations promptly.