ChatGPT developers are quietly developing new artificial intelligence - Reuters

According to Reuters, OpenAI, the creator of the virtual assistant ChatGPT, is working on a new approach to intelligent technology.

As part of the project, nicknamed "Strawberry," the company supporting Microsoft is trying to improve the efficiency of its model, the agency said in a press release on Friday.

A person familiar with the matter told Reuters that even within OpenAI, Strawberry's working methods are "top secret."

The plan includes a "special way" of pre-training and post-processing artificial intelligence models on large data sets, the sources said. The goal, the sources explain, is to allow artificial intelligence to generate answers to questions, but it is necessary to plan in advance to conduct what is called "deep research" through the waves online in a safe and reliable way. Reuters said it has reviewed internal OpenAI documents detailing the US company's plans to use Strawberry for research. However, the agency said it could not yet determine when the technology would be available to the public. Sources are intended as "work in progress."

To address this issue, an OpenAI spokesperson told Reuters: "We want our artificial intelligence models to see and understand the world as we (humans) do." It is common practice in the industry to continuously research new artificial intelligence capabilities, and it is widely believed that the cognitive capabilities of these systems will improve over time.

Today's advanced language learners can summarize large amounts of text and organize speech faster than humans, but often struggle with solutions that are meaningful and meaningful. to people. When this happens, students often create "falsifications" to present false or misleading information.

Researchers interviewed by Reuters say thinking skills that artificial intelligence models can't master are key to reaching human levels or higher.

Last week, Yoshua Bengio, one of the world's leading artificial intelligence experts and a pioneer in deep learning, also warned of the "many problems" private companies could face, including "death man." It reaches human-level or higher artificial intelligence.

"Entities are smarter than people, they have a goal: can we make sure they work on our lives?" wrote a professor at the University of Montreal and scientific director of the Montreal Institute for Learning Algorithms (MILA) in a web article.