- Google and OpenAI are not positioned to win the AI arms race, a Google engineer said in a leaked doc.
- A document written by a senior Google engineer suggests the open-source community is rivaling them.
- The document published by SemiAnalysis said Google had “no secret sauce” to help them succeed.
Google and OpenAI are not positioned to win a hotly contested AI arms race as the open-source community is “lapping” them, a senior engineer at the search giant has said.
A leaked internal document, first shared by an anonymous account on a Discord server and then published by consulting firm SemiAnalysis after verifying its authenticity, said Google had “no secret sauce” in an AI race that has taken shape in recent months.
The document, written by Google senior software engineer Luke Sernau, according to Bloomberg, suggested that although Google has “done a lot of looking over” its shoulders at rival OpenAI, open source engineers have been quietly surpassing them both.
“The uncomfortable truth is, we aren’t positioned to win this arms race and neither is OpenAI. While we’ve been squabbling, a third faction has been quietly eating our lunch,” the document noted. “I’m talking, of course, about open source. Plainly put, they are lapping us.”
Sernau added that without “secret sauce,” Google’s best hope was “to learn from and collaborate with what others are doing outside Google,” and prioritize third-party integrations.
Sernau said open-source engineers were doing things with $100 that “we struggle with” at $10 million, “doing so in weeks, not months.”
The comments come amid a fierce new race for AI supremacy, which has been underway since the release of ChatGPT in November, with both Google and Microsoft-backed OpenAI attempting to keep pace with each other on advancements in their AI models.
Though Sernau said that “our models still hold a slight edge in terms of quality,” open-source engineers have managed to solve several things that the likes of Google still consider “major open problems.”
He cited challenges around running AI models on a phone as one such major issue, which he said open-source engineers have managed to do on a Google Pixel 6. “Open-source models are faster, more customizable, more private, and pound-for-pound more capable,” he wrote.
OpenAI, and Sernau did not immediately respond to Insider’s request for comment made outside of normal working hours. Google declined to comment on the record.