Meta says it will begin rolling out Llama 3 in the next few weeks, while Microsoft-backed OpenAI says the GPT-5 version will be “coming soon”.
"We're working to find ways to make these models not only able to talk, but also to actually reason, plan, and have memory," said Joelle Pineau, vice president of AI research at Meta.
OpenAI CEO Brad Lightcap told the FT in an interview that the next generation of GPT will show progress in solving “difficult problems” such as reasoning. “We will begin to see AI able to take on more complex tasks in more complex ways.”
Lightcap added that while today's AI systems "perform small, one-time tasks very well," their capabilities are still only being exploited in a "fairly narrow" area.
The upgrades to Meta and OpenAI are part of a wave of new big language models being released this year by companies including Google, Anthropic, and Cohere.
Reasoning and planning are crucial steps toward what AI researchers call “generalized artificial intelligence”—human-level cognition that enables chatbots and virtual assistants to complete related task sequences and predict the outcomes of actions.
Yann LeCun, head of Meta's AI research team, says current AI systems produce results by piecing together words without actually "thinking and planning." They struggle with complex questions or remembering information for extended periods, leading to "stupid mistakes."
Meta plans to embed its new AI model into WhatsApp and Ray-Ban smart glasses. They are preparing to release Llama 3 in multiple sizes, for different apps and devices, in the coming months.
Chris Cox, Product Director at Meta, shared an example of Llama 3's capabilities when integrated into smart glasses, such as guiding the wearer through coffee machine repairs.
Source
Comment (0)