A OpenAI, creator of ChatGPT, is working on a new approach to its artificial intelligence (AI) in a project called “Strawberry“, according to a source familiar with the matter and internal documentation reviewed by Reuters.
ADVERTISING
This project, whose details had never been disclosed, comes at a time when the startup supported by Microsoft runs to show that its models are capable of offering advanced reasoning capabilities.
Teams within OpenAI are working on Strawberry, according to a copy of a recent internal document seen by Reuters in May. The news agency was unable to determine the precise date of the document, which details a plan for how the OpenAI intends to use Strawberry to conduct research. The source described the plan to Reuters as a work in progress. Reuters was also unable to find out how close Strawberry is to being publicly available.
How Strawberry works is a well-kept secret, even within the OpenAI, said the source.
ADVERTISING
The document describes a project that uses Strawberry models with the goal of allowing the company's AI to not only generate answers to questions, but also plan enough to navigate the internet autonomously and reliably to accomplish what the company wants. OpenAI calls “deep research”.
This is something that has eluded AI models to date, according to interviews with more than a dozen AI researchers.
Questioninformation about Strawberry and the details reported in this story, a spokesperson for OpenAI said in a statement: “We want our AI models to see and understand the world more like we do. Continuous research into new AI capabilities is a common practice in the industry, with the shared belief that these systems will improve your thinking over time.”
ADVERTISING
The spokesperson did not directly respond to questions about Strawberry.
The Strawberry project was previously known as Q*, which Reuters reported last year was already seen within the company as a step forward.
Two sources described seeing earlier this year what employees at OpenAI they were told they were demonstrations of Q*, capable of answering complicated science and mathematics questions beyond the reach of commercial models available today.
ADVERTISING
At an internal meeting on Tuesday, the OpenAI showed a demonstration of a research project that she said had new human-like reasoning abilities, according to the Bloomberg. A spokesperson for OpenAI confirmed the meeting, but declined to give details of its contents. Reuters was unable to determine whether the design demonstrated was Strawberry.
A OpenAI expects the innovation to dramatically improve the reasoning capabilities of its AI models, the person familiar with the matter said, adding that Strawberry involves a specialized way of processing an AI model after it has been pre-trained on very large data sets. big ones.
Researchers interviewed by Reuters say reasoning is the key to AI achieving human or superhuman intelligence.
ADVERTISING
While large language models can now summarize complex texts and compose elegant prose much faster than any human, the technology often fails on common-sense problems whose solutions seem intuitive to people, such as recognizing logical fallacies and playing dice games. old. When the model encounters this type of problem, it often “hallucinates” false information.
AI researchers interviewed by Reuters generally agree that reasoning, in the context of AI, involves forming a model that allows the AI to plan for the future, reflect on how the physical world works, and work on challenging multi-step problems reliably.
Improving reasoning in AI models is seen as the key to unlocking the models' ability to do everything from make major scientific discoveries to plan and build new software applications.
The CEO of OpenAI, Sam Altman, said earlier this year that in AI “the most important areas of progress will be around the ability to reason.”
Other companies like Google, Meta e Microsoft They are also experimenting with different techniques to improve reasoning in AI models, as are most academic labs carrying out AI research. However, researchers differ on whether large language models (LLMs) are able to incorporate long-term ideas and planning into their forecasting. For example, one of the pioneers of modern AI, Yann LeCun, who works at Meta, has often said that LLMs are not capable of human reasoning.
Strawberry is a key component of the OpenAI to overcome these challenges.
Read also
