Meta announced on February 24, the development of an AI for Facebook. It is a system inspired by ChatGPT technology, with features to generate texts, chat and analyze data. According to Mark Zuckerberg, CEO of the company, the technology it is still in the “state of the art”. That is, a kind of testing phase. It is not yet open to the general public, so access is only for developers.
see more
Google develops AI tool to help journalists in…
Unopened original 2007 iPhone sells for nearly $200,000; know...
In the face of advances in the field of artificial intelligence, several companies in the technology sector have sought their own systems to compete with each other. With the arrival of ChatGPT, an AI program capable of performing various tasks such as writing, translating and analyzing data, other similar programs were created.
Google, in this wave, introduced the Bard. Of course, Microsoft also invested in OpenAI, the company that created ChatGPT, whose result was integrated into Bing. Not to mention some Chinese projects that are on the way.
LLaMa
Following the wave, Facebook announced the Large Language Model Meta AI (LLaMa), an AI with functionality similar to ChatGPT. So far, the technology is restricted to language specialist developers. The current model of the program has the 20 most spoken languages in the world, with the Latin alphabet.
LLaMa's target audience
The platform's AI will be offered to researchers, laboratories, institutes and NGOs working with language artificial intelligence. LLaMa is available in four language models and will make the system open to the scientific community and research.
It is speculated that this technology could be used to combat language programs that use prejudiced or discriminatory phrases. In addition, it could be used to combat fake news, so that it will be able to cross-reference news data and be able to identify false points or misinformation.
According to Meta: “As a basic model, the LLaMA was designed to be versatile and can be applied to many different use cases compared to a model developed for one task specific".
