According to a Wall Street Journal story, Meta is creating an AI model to rival GPT-4.
Compared to Llama 2, the open-source AI that Meta recently released, it is anticipated to be significantly more potent.
The action indicates that the AI weapons race is intensifying as Meta tries to overtake competitors.
In order to compete with OpenAI’s most advanced model, Meta is creating an AI.
Unnamed individuals told The Wall Street Journal that the tech giant was developing a new AI system that intended to be as potent as GPT-4, the most recent iteration of OpenAI’s large language model.
The announcement comes as Big Tech businesses are rushing to create their own AI models in an effort to catch up to the company’s competitors.
The model, which is anticipated to be substantially more potent than the business’s recently announced Llama 2 AI, could be made available by Meta at some time in 2019.
According to reports, corporations would use the AI largely as a tool for text generation and analysis. This is comparable to ChatGPT enterprise, the hugely successful chatbot from OpenAI that debuted last month for businesses.
Unnamed sources told the Journal that Meta CEO Mark Zuckerberg was pushing for the future AI to be open source, breaking with similar models introduced by other major tech companies like ChatGPT and Google’s Bard.
Although Meta’s models have limitations and users are required to abide by the company’s terms of service, open-source models are accessible and allow users to tweak and utilize the AI as they like.
Despite how adaptable open-source methods are, some people worry that they might be abused for bad things like disseminating false information or violating intellectual property rights. According to The Journal, Meta’s lawyers expressed alarm over possible abuses of the company’s AI model.
As interest in the developing technology has increased, Meta has progressively turned its attention to it; in a recent earnings call, Zuckerberg described it as a significant theme for the firm.
However, internal memos from last year obtained by Reuters reveal that the corporation has a “significant gap” in AI research and investment, making it difficult for it to keep up with OpenAI and Microsoft.
Since then, the Facebook owner has been making quick purchases of the expensive H100 chips required to train huge language models, and it is anticipated that next year it will rank second globally in terms of its H100 holdings.
To cut the lead amassed by its competitors, Meta has invested on opening-source its AI models. Earlier this year, the business declared it will open source its Llama 2 model, making it available for free for both study and commercial use.