Mark Zuckerberg has confirmed that Your company wants to create general AI (AGI) with the apparent objective of competing against OpenAI and other artificial intelligence companies that are also working on this technology. The CEO of Meta, yes, has detailed in an interview with The Verge which does not have an exact definition about general AI, and does not know when they will reach it.
“We have come to the conclusion that, to create the products we want, we need to create general intelligence,” Zuckerberg has highlighted, with the aim of recruiting AI researchers who are interested in working on this technology. “I think it’s important to convey that because many of the best researchers want to work on the most ambitious problems.”
In addition to searching for new talent to help create an overall AI model, Meta is also acquiring a large number of components to work on the development of this technology. Zuckerberg, in fact, has highlighted that his company will have a total of 600,000 Nvidia GPUs by the end of 2024 intended for AI work. “We have developed the capacity to do this on a scale that may be greater than that of any other individual company,” he told the aforementioned media.
The CEO of Meta is not clear about what general AI is, nor when it will be available
Mark Zuckerberg, yes, It does not have a clear definition about what artificial general intelligence is and what it can do.. He says that “one can argue whether general intelligence is similar to human-level intelligence, or is like a ‘human-plus’, or is some super intelligence from the distant future.” For him, however, what is important is “its breadth, which is that intelligence has all these different capacities where you have to be able to reason and have intuition,” she says.
The CEO of Meta has also spoken about the arrival of general AI. Again, he’s not sure when we’ll see her. In fact, he claims that It is not something that will come in a single moment, but rather it will be something gradual. It is precisely something that we have seen with other types of AI, such as generative AI, which has advanced considerably in recent years.
Meanwhile, Meta continues working on Call 3its AI model that will replace Llama 2 in the future and that includes a very important novelty: the ability to generate code.
Curiously, and unlike GPT, the current Meta language model is not capable of generating code because Meta did not see the coding as something relevant or something that the user used. “It’s not like many people are going to ask questions about coding on WhatsApp,” she said. She was wrong. He himself has confirmed that “coding is structurally very important so that LLMs can understand the rigor and hierarchical structure of knowledge and, in general, have a more intuitive sense of logic.”
An open source AI. Or maybe closed
In the interview with The Verge, Zuckerberg also spoke about whether make your general AI an open source model, allowing third parties to use it, or closed source, only for Meta. The executive sees drawbacks in both cases. A closed AI is more valuable, but at the same time it would end up “making everything more concrete.” With an open one, on the other hand, “it addresses a broad class of problems that could arise from unequal access to opportunity and value,” but it could also make the use of AI more dangerous.
The CEO of Meta says that, “as long as it makes sense and is the safest and most responsible thing to do,” will lean towards open source. He emphasizes, however, that he does not want to be forced to do something because he said he would do it, referring to these words.