OpenAI, Meta to roll out AI models capable of ‘reasoning’

OpenAI

Microsoft-backed OpenAI and Meta will soon release artificial intelligence (AI) models that will be able to reason and plan, a report has said.

While executives of Meta said they will be rolling out Llama 3 in the coming weeks, OpenAI said its next version of AI model "GPT-5" will be launched soon, London-based Financial Times newspaper has reported.

Joelle Pineau, vice-president of AI research at Meta, told the publication that the company is figuring out how these large language models can reason, plan, and have memory.

Large language models power generative AI applications such as hugely popular ChatGPT and Llama.

OpenAI’s chief operating officer Brad Lightcap said that the next version of GPT would solve hard problems such as reasoning.

Lightcap said today's AI systems are just scratching the surface and have narrower capabilities. AI will be able to solve more complex tasks in a sophisticated manner in the future, he said.

The AI race is heating up with new planned launches by rivals including Google and Anthropic.

Meta intends to integrate its new AI model into WhatsApp and Ray-Ban smart glasses, the report added. It is gearing up to unveil Llama 3 in various model sizes, tailored for diverse applications and devices, within the upcoming months.

Also read: OpenAI may have 'scraped' data from YouTube videosMeta to start labelling AI-generated images from other platforms