A senior software engineer at Google wrote a critique stating that the internet search leader has lost its share of artificial intelligence in the open-source community, where many independent researchers are using AI technology to create fast and unexpected development.
The engineer, Luke Sernau, published the document in an internal Google system in early April. In the past few weeks, the document has been shared thousands of times by Googlers, according to a person familiar with the matter, who asked not to be named because they are not authorized to discuss internal company matters. On Thursday, the document was published by the consulting firm SemiAnalysisand making the rounds in Silicon Valley.
In Sernau’s analysis, Google’s rivalry with the OpenAI startup has distracted from the rapid developments made in open-source technology. “We’ve been looking over our shoulders a lot at OpenAI. Who’s going to cross the next milestone? What’s the next step?” he wrote. “But the uncomfortable truth is, we are in no position to win this arms race and neither is OpenAI. While we were fighting, the third faction quietly ate our lunch. I am talking, of course, about open source.
Sernau did not respond to a request for comment.
As the development of generative artificial intelligence gathers pace, employees at Google and other tech giants are engaged in lively discussions internally and externally about the technology that is changing their industry. Google, in particular, has under pressure because the wild popularity of OpenAI’s chatbot ChatGPT has sparked concerns that the company may lose its edge in artificial intelligence, a field in which it has been around for a long time. became a leader.
But Sernau stated that the real threat to Google comes from open-source communities, where engineers are rapidly developing models that rival the quality of major technology companies, and can be cheaper. These models, he said, will be faster, more adaptable and more profitable than Google’s.
“We don’t have a secret sauce,” Sernau wrote. “Our best hope is to learn from and collaborate with what others besides Google are doing.” He expressed concern that clients are not willing to pay for models with high-quality technology offered for free.
A Google spokesperson did not comment on the content of the post. In a recent earnings call, Alphabet’s Chief Executive Officer Sundar Pichai said, “Our investments and achievements in AI over the last decade have positioned us well,” pointing to the development progress of models and work with developers and other partners. Pichai called AI regulation beforewhich warns that the technology can be “very harmful,” if not used in a thoughtful way.
In February, a large language model developed by Meta was leaked to the open, generative AI development startup on open-source forums. The model, known as LLaMA, smaller than the models preferred by Google and OpenAI, which makes it easier to work with; Researchers currently must apply to Meta to access LLaMA.
Google would do well to shift its focus to smaller models, Sernau argued. “Giant models slow us down,” the engineer wrote. “In the long run, the best models are the ones that can be changed quickly.”