OpenAI’s mysterious project “Strawberry” finally surfaced! Previously, a picture of strawberries posted by OpenAI CEO Sam Altman on social media triggered widespread speculation. Now, The Information reports that OpenAI has presented the secret AI project to U.S. national security officials and assured that its development and use will be safe and sound. This article will delve into Strawberry’s features, application prospects, and impact on the future development of OpenAI.
You may still remember that OpenAI CEO Sam Altman posted a photo of strawberries on social media at the beginning of this month, sparking speculation that the AI model "Strawberry" will be released.
Now, there is finally new news about "Strawberry"! According to The Information, OpenAI has shown the secret strawberry project to U.S. national security officials and ensured that it will be developed and used within a safe and reasonable scope.
According to insiders, this AI product is expected to be available to everyone this fall. It will be part of the chatbot and may be built into ChatGPT. Where Strawberry shines is its ability to solve math problems, which is rare among current chatbots. Not only that, it can handle programming problems, but its applications extend far beyond that.
After getting more "thinking" time, the Strawberry model can even answer more subjective questions about product marketing strategies. OpenAI employees have demonstrated Strawberry's ability to solve the New York Times Connect word puzzle to colleagues, demonstrating its superiority at language tasks.
OpenAI's efforts to launch Strawberry are part of its efforts to stay ahead of the competition in the conversational artificial intelligence field. This technology has far-reaching implications for future products, such as multi-step tasking agents, which are expected to open up additional revenue streams for the company.
OpenAI's business is growing at an impressive rate, with its LLM and ChatGPT subscription revenue tripling year-on-year to $283 million per month. Still, with the company privately valued at $86 billion, its prospects largely depend on the successful launch of its new flagship large language model (LLM), codenamed Orion.
Strawberry’s version of the chatbot, the Distilled Edition, is designed to maintain the same performance as the original model while being easier to operate and less expensive. In addition, OpenAI also uses a larger version of Strawberry to generate the data for training Orion. This AI-generated data is called "synthetic data" and helps the company overcome the limitations of obtaining high-quality data.
Using the high-quality training data generated by Strawberry can help reduce the errors produced by the model, so-called "illusions". Imagine a model without illusions that could accurately answer logical puzzles. That would be a huge improvement.
Highlight:
?OpenAI’s upcoming new AI focuses on solving complex mathematical and programming problems.
OpenAI’s LLM and ChatGPT subscription revenue tripled year-on-year, with the company valued at US$86 billion.
Using the high-quality training data generated by Strawberry, it is expected to significantly reduce the errors produced by AI models.
All in all, the emergence of Strawberry not only represents another breakthrough in AI technology for OpenAI, but also indicates that future AI models will make greater progress in solving complex problems, improving efficiency, and reducing costs. It will have a profound impact on OpenAI and even the entire AI industry, and deserves our continued attention.