No Downside to Sharing AI: Meta’s Insightful Take
During an event on Thursday, executives from Meta Platforms Inc. stated that openly sharing its AI technology has not resulted in any significant disadvantages, despite the contrary approach taken by many other companies.
Over the past few months, Meta has released open source versions of its large language models – the technology behind AI chatbots like ChatGPT. The idea is to keep those models free and then gain an advantage by building products and services on top of them, executives said at the FAIR event at the company’s artificial intelligence research lab.
“There’s really no commercial harm if it’s also available to other people,” said Yann LeCun, Meta’s lead AI researcher.
Meta has joined most of the world’s biggest tech companies in embracing generative AI that can create text, images, and even videos based on simple prompts. But they don’t walk the same path. Many of the top AI developers, including OpenAI and Google’s DeepMind, are not currently open-sourcing their major language models.
But staying open has its advantages. Meta can count on thousands of developers around the world to help improve their AI models. This means that the technology could potentially develop faster than if the company only used its own researchers.
“If you do this in secret, you’re going to be left behind,” LeCun said.
The recent turmoil at ChatGPT creator OpenAI has fueled a global debate about the evolution of artificial intelligence and whether companies should be more open in their approach to developing potentially game-changing technology.
Over the summer, Meta released commercial versions of Llama 2 and Code Llama, an artificial intelligence technology that competes with ChatGPT and Microsoft Corp.’s GitHub Copilot. By offering free access to its large language models that require massive amounts of computing power, Meta allows smaller companies and developers to build AI tools.
Meta plans to charge some big companies, such as Microsoft, Amazon.com Inc. or Alphabet Inc.’s Google, for using its large language models, CEO Mark Zuckerberg told investors earlier this year. “I don’t think it’s going to be a huge amount of revenue in the near future, but in the long run, hopefully it can be something,” he said.
Meta’s FAIR team, which stands for Fundamental Artificial Intelligence Research, was founded ten years ago and now has more than 400 researchers in North America and Europe. Recently, the team has focused more on finding product applications for their research. In January, Meta created a new artificial intelligence product group that focuses exclusively on producing generative artificial intelligence technologies for its products.
