Meta developing new powerful generative AI system

According to a Wall Street Journal (WSJ) report, Meta Platforms is currently developing a new AI system. Facebook and Instagram’s parent company wants the new AI model to be as capable as OpenAI’s most sophisticated system and to be more potent than its current open-source ‘Llama 2’. Released in July and distributed by Microsoft’s Azure cloud, Llama 2 was supposed to be an alternative to OpenAI’s ChatGPT and Google’s Bard.

Why does it matter?

Since OpenAI launched ChatGPT in November last year, the generative AI market has witnessed several companies compete for more innovative capabilities and upgraded processes. According to the WSJ, Meta CEO Mark Zuckerberg assembled a specific team to speed up the development of GenAI. The new large language model (LLM)-based system will be trained later this year and launched in 2024. Meta’s upcoming AI model is expected to be a serious challenger in the GenAI global race. Apple is also rumoured to be working on products comparable to ChatGPT and Bard. Apple’s effort to create an ‘Apple GPT‘ chatbot relies on ‘Ajax’, its proprietary foundation for building LLMs.

Furthermore, Zuckerberg’s statement also underlined that the AI tool would be open-source, which means anyone can look under the bonnet and evaluate its algorithms.

The issues with Meta’s open-source models or approaches have been questioned in recent months. The Open Source Initiative (OSI), a non-profit organisation dedicated to upholding open-source software standards, has recently criticised another Meta project, Code Llama, for not adhering to conventional open-source software licenses.