Meta headquarters

Meta develops new AI model to rival OpenAI’s ChatGPT

Meta (NASDAQ: META) is reportedly working on a brand new artificial intelligence (AI) model designed to rival OpenAI’s ChatGPT barely weeks after the commercial release of its large language model, Llama 2.

According to a report by the Wall Street Journal (WSJ), Meta‘s new model is expected to compete favorably with the latest iteration of ChatGPT and will begin training in early 2024. WSJ says the incoming model will most likely offer the same functionalities of generative AI but with the addition of additional parameters and robust training data.

Unlike ChatGPT, Meta is expected to make its proposed AI model open-source and free for companies to leverage the offering to create AI tools.

Meta has reportedly put things in motion by assembling a team of AI experts since the start of the year. While Meta’s first foray into AI relied on Microsoft’s Azure (NASDAQ: MSFT), the company is keen on building its native infrastructure to train future AI models.

The company has been bolstering its reserves with Nvidia’s H100 chips, tailored explicitly for training LLMs. However, while WSJ’s report did not disclose the extent of Meta purchases, insiders say the company may have splurged millions of dollars on the latest chips.

Despite the investment in personnel and infrastructure, the company faces stiff competition from industry leaders as they continue to push the frontiers for AI innovation.

Meta must catch up with the pecking order regarding AI development following its late entry into the space. The company is currently playing catchup with OpenAI and Google (NASDAQ: GOOGL), given the successes of their generative AI platforms due to their first mover advantage.

Meta released the commercial version of its open-source LLM dubbed Llama 2 to close the gap, opting to lure institutional and retail users via a free offering. In the following weeks, the tech giant hinted at a series of AI offerings, including a tool to assist developers “using text prompts to generate and discuss code.”

“Meta realized they were behind on the currency AI hype cycle, and this gives them a way to open up the ecosystem and seem like they are doing the right thing, being charitable and giving back to the community,” said an unnamed source.

Meta future in the AI space

While Meta’s support for open-source AI systems has elicited praise in certain quarters, critics say there are gaping flaws that threaten the safety of such models. They argue that open-source AI models may increase the risk of misinformation or other forms of misuse since they are open to a wide range of users.

Meta has been enmeshed in a class action lawsuit bordering on the illegal use of personal data to train its AI models as authorities flex their regulatory muscles on emerging technology to protect consumers.

Watch: AI, ChatGPT, and Blockchain | CoinGeek Roundtable with Joshua Henslee

YouTube video

New to blockchain? Check out CoinGeek’s Blockchain for Beginners section, the ultimate resource guide to learn more about blockchain technology.