Meta has been snapping up AI coaching chips and constructing out information facilities to be able to create a extra highly effective new chatbot it hopes will likely be as refined as OpenAI’s GPT-4, in line with The Wall Street Journal. The corporate reportedly plans to start coaching the brand new massive language mannequin early in 2024, with CEO Mark Zuckerberg evidently pushing for it to once again be free for firms to create AI instruments with.
The Journal writes that Meta has been shopping for extra Nvidia H100 AI-training chips and is beefing up its infrastructure in order that, this time round, it received’t have to depend on Microsoft’s Azure cloud platform to coach the brand new chatbot. The corporate reportedly assembled a bunch earlier this 12 months to construct the mannequin, with the objective of dashing up the creation of AI instruments that may emulate human expressions.
That objective appears like a pure extension of rumored generative AI options Meta has already been engaged on. A June leak claimed there was an Instagram chatbot with 30 personalities being examined, which sounds quite a bit just like the unannounced AI “personas” the corporate is said to be launching this month.