Meta Platforms, Facebook’s parent company, is working on a new artificial intelligence system that aims to be as advanced as OpenAI’s top model – GPT-4, according to The Wall Street Journal. The upcoming AI model is expected to launch next year and to be much more powerful than their current model – Llama 2.
Llama 2, an open-source AI language model from Meta, was launched in July and is available through Microsoft’s Azure cloud services. It’s designed to compete with OpenAI’s ChatGPT and Google’s Bard.
Meta is currently developing a new system, the details of which might change. This system is expected to help other companies create services that can generate advanced text, perform analyses, and produce other outputs.
The Journal reports that Meta is buying more Nvidia H100 AI-training chips and boosting its infrastructure. This is to avoid relying on Microsoft’s Azure cloud platform for training the upcoming chatbot. Earlier this year, the company formed a team to speed up the creation of AI tools that can mimic human expressions.
Since OpenAI’s ChatGPT was launched last year, the generative AI market has attracted a lot of attention from companies. They’re looking to use these technologies to improve their capabilities and make their operations more efficient.
Apple has created its own system, ‘Ajax’, for building large language models and is reportedly testing a chatbot unofficially called ‘Apple GPT’, according to some project engineers. Google and Microsoft have also been integrating more AI into their productivity tools, with Google aiming to include generative AI in Google Assistant. Amazon is also working on generative AI projects across its company, which could lead to an Alexa powered by a chatbot.