OpenAI and Meta Reportedly Preparing New AI Models Capable of Reasoning

Soon?

OpenAI and Meta are on the verge of releasing next versions of their AI models that will supposedly be capable of reasoning and planning, the Financial Times reports. But, as with any hype coming out of big tech, take it all with a grain of salt.

For one thing, there's no exact timeline for a rollout. Per the report, OpenAI has indicated that its next AI model, which will likely be called GPT-5, is coming "soon." Meta, meanwhile, said it's expecting to release its Llama 3 model in the coming weeks.

"We are hard at work in figuring out how to get these models not just to talk, but actually to reason, to plan... to have memory," Joelle Pineau, Meta's vice president of AI research, told the FT.

If either party's claims are to be believed — a big if, considering how superficial they sound — both models could mark a major step towards achieving artificial general intelligence (AGI), an eagerly-fantasized breakthrough that could see AI models possessing human-like or even superhuman cognition.

Pinky Promise

Brad Lightcap, the chief operating officer at OpenAI, told the FT that "we're going to start to see AI that can take on more complex tasks in a more sophisticated way."

"I think we're just starting to scratch the surface on the ability that these models have to reason," he added.

With its upcoming Llama 3, Meta aims to integrate a more powerful version of its AI assistant into WhatsApp, as well as its augmented reality Ray-Ban glasses, according to the report.

OpenAI's plans for GPT-5 are less clear.

"I think over time... we'll see the models go toward longer, kind of more complex tasks," Lightcap told the FT. "And that implicitly requires the improvement in their ability to reason."

AGI... Someday

The claims coming out of both camps are frustratingly vague. Of course, that's more or less been the modus operandi for AI leaders. After all, this isn't the first time we've been teased with glimmers of sentient AI.

Last summer, researchers at Microsoft claimed that GPT-4, which at the time had only just been released, was showing "sparks" of AGI. OpenAI CEO Sam Altman has long spoken of the company's AGI endeavors. At Meta, CEO Mark Zuckerberg has made it clear that creating AGI is the driving vision behind its expensive AI push.

Some research has suggested that evidence of an AGI breakthrough are nothing more than a "mirage" or illusion. Broadly, there's a lot of skepticism surrounding the idea — perhaps now more than ever. At some point, AI models may simply run out of data to train on.

By contrast, other experts, including AI "godfather" Geoffrey Hinton, expect AGI as soon as within the next few years.

However rigorous those expert projections may be, these latest promises from OpenAI and Meta still sound more like gossip than anything substantive, at least until we've seen them with our own eyes.

More on AI: New Law Would Force AI Companies to Reveal Copyrighted Training Data