Nick Clegg, Meta’s president of global affairs, is back with another opinion piece on artificial intelligence. The executive and former British politician who recently promoted the metaverse narrative published an article in the Financial Times today (July 11) advocating for open-sourced large language models (LLM).
“It’s not sustainable to keep foundational technology in the hands of just a few large corporations,” Clegg said in the post.
Meta developed an LLM with the capability of powering chatbots earlier this year, similar to OpenAI’s ChatGPT and Google’s Bard. But unlike OpenAI and Google, the Facebook and Instagram parent made the underlying code available for download, meaning it can be copied and modified to create new chatbots. According to Meta, open-sourcing its technology will democratize access to AI among researchers and help them test bias and toxicity within LLMs. Critics worry the technology could be manipulated by bad actors to spread disinformation and hate speech. Days after Meta launched its LLM with the intent of vetting the researchers and academics who earned access to the model, it leaked on 4chan, a messaging platform known for spreading misinformation.
Meta isn’t the only company to open its LLM code—so have Abu Dhabi’s Technology Innovation Institute and San Francisco’s MosaicML—but it might be the biggest company to do so. While the decision gave Meta a competitive advantage in the AI race, it also put the company under a microscope. It is Clegg’s job to justify the business move and convince users that Meta has their best interests in mind.
Clegg previously spent 12 years in the U.K. parliament, as the leader of the Liberal Democrat party and as deputy prime minister. He joined Meta in 2018 as a lobbyist and communications officer. CEO Mark Zuckerberg promoted him last year to lead Meta’s policy, which includes publicly advocating for the company’s products and work. The Financial Times opinion piece is the third post from Clegg since April promoting Meta’s AI and metaverse technologies, which both operate under the company’s Reality Labs division. Clegg previously wrote about how the metaverse can aid in classroom learning and clarified how Meta’s algorithms work.
In Clegg’s newest article, he makes the case that openness is the best solution for settling fears around AI, and it can be aided by industry collaboration, stress testing the technology and sharing developments through academic papers or public announcements. Open-sourced code is something that already runs the internet, web browsers and apps, he said, so it isn’t something that should be cause for concern. Mozilla Firefox, Signal and WordPress are all free, open-sourced software.
“Openness isn’t altruism — Meta believes it’s in its interest,” Clegg wrote. “It leads to better products, faster innovation and a flourishing market, which benefits us as it does many others.”
