Meta Reportedly Working Towards Powerful AI System

Meta is set on developing on a new artificial intelligence system that will be as powerful as the most advanced model by OpenAI.

Meta Wants More Power than Llama 2

Last July, Meta introduced Llama 2 which is in partnership with Microsoft. However, The Wall Street Journal reports that Meta is gearing up to take it to the next level. The new AI model is targeted to launch some time next year and is on its way to be several times more powerful than Llama 2.

The report also showed that this is Meta's way of asserting their place as a major player in the AI world after falling behind its rivals. In 2023 alone, several companies have been improving their AI models triggering a tight competition in the industry. Moreover, AI regulations are also being prioritized due to its fast progression.

Meta CEO Mark Zuckerberg is set to attend an AI forum hosted by Senate Majority Leader Chuck Schumer, along with other tech executives this September 13.

Here's What We Know About Meta's Plans

Currently, the company is reportedly acquiring the most advanced Nvidia chips used for AI training, the H100s. In addition, data centers are being built to support the training. Other details could still change along its way but Meta expects to start the training in early 2024.

While the model might not still be in close competition with other leaders in AI, Meta still hopes that its new upcoming AI model will be as capable as OpenAI's GPT-4. On the other hand, Meta seeks advantage on its open-sourced approach as it can be more adaptable and less expensive.

However, open-source model can be risky due to its legal concerns. Meta's lawyers have also raised some of these concerns over the new plan.