In the secret rooms of OpenAI, the now famous Artificial Intelligence company and parent of ChatGPT, is developing a secret project, known as Strawberry, destined to revolutionize the technological landscape of AI. Second Reuters, who was able to view internal company documents, Strawberry would represent a new step towards general artificial intelligence (AGI), with the ability to reasoning advanced that yes inspire human potential.
According to rumors, Strawberry would be none other than an evolution of Q*, the top secret OpenAI project that was talked about in November during the tumultuous period of push and pull between the board of directors and Sam Altman's return to lead the company. Various sources have confirmed that demonstrations of Q* were shown earlier this year showing the ability to answer extremely complex scientific and mathematical questions, overcoming the limits of the models currently available on the market. Despite these revelations, the project Strawberry, for now, remains shrouded in secrecy, with the management of the San Francisco-based company determined to maintain a strict secrecy about this latest innovation.
Strawberry will browse the internet independently
Strawberry aims to equip AI models with advanced autonomous capabilities. These models will not only be able to answer complex questions, but they can also anticipate and prepare future questions, with the assistance of a “CUA” (computer-using agent), browsing the Internet independently. This capability is seen as a significant leap towards artificial general intelligence (AGI), allowing AIs to explore complex information autonomously and reliably on the web and plan activities without the need for external resources. This approach, defined “deep research” from OpenAI, represents a new achievement in the field of artificial intelligence, similar to human vision in understanding and navigating the world.
OpenAI thus aims to get ahead of everyone in the field of prediction, an area that has so far been little explored by large artificial intelligence companies, a bit like they did with ChatGPT last year. Already during the year Altman had declared that the main focus of OpenAi was to improve the reasoning capacity of AI models. This ability is crucial not only for solving complex problems, but also for understanding and interacting more naturally with the surrounding environment.
AI: Strawberry will improve himself
Strawberry, inspired by the method “Self-Taught Reasoner” (STaR) developed at Stanford in 2022, aims to enable AI models to improve yourself independentlyand analyzing their responses and the generation process. Although still in development, Strawberry represents a clear signal of OpenAI's commitment to maintaining leadership in the field of artificial intelligence by continually challenging its competitors, according to Reuters.
“We want our AI models to see and understand the world more like us. Continued research into new AI capabilities is common practice in the industry, and we believe these systems will refine their reasoning capabilities over time,” an OpenAI spokesperson said.
Despite the concerns, OpenAi is always aiming higher
Strawberry's game-changing potential, while promising, comes with series ethical and safety concerns. Industry experts raise doubts about ethics and on the responsible management of access to such advanced technologies.
Despite this, OpenAI continues to forge ahead while maintaining leadership in the artificial intelligence sector with projects like Strawberry, which promise to push the boundaries of AI towards a future where this technology not only assists but interacts and learns in ways increasingly similar to human beings, making judgments, reasoning, communicating in natural language, and learning from the surrounding environment. The hope remains that such projects will never turn into something like Skynet.
