Its top developer Google wrote an internal memo to the company, warning that the tech giant was losing its research edge. Artificial intelligence, and the primary is now being contested by the “open source” community, where many independent researchers use AI. making rapid and largely unexpected progress.

Software engineer Luke Chernau shared his concerns on Google’s internal platform in early April, and in recent weeks his text has been shared with Google employees thousands of times, according to a source familiar with the matter, and he spoke to Bloomberg without being disclosed as it is not authorized. discuss the internal affairs of the company.

On Thursday, the article was published by the consulting firm SemiAnalysis and was distributed throughout Silicon Valley..

According to Chernau’s analysis, Google’s competition with billion-dollar Microsoft-funded startup OpenAI has diverted attention from the rapid development of open source technology. “We watched a lot. behind our backs, in OpenAI. Who will reach the next milestone? What will be the next move? But the inconvenient truth is that we are not in a position to win this race, nor is OpenAI. What about us we quarreled, someone else “quietly ate our food”. I’m talking, of course, about open source.”

Chernau did not respond to Bloomberg’s request for comment.

As the pace of artificial intelligence accelerates, employees at Google and other tech giants are speaking loudly, privately and publicly about technologies that are changing their industry. Google and Microsoft have been heavily criticized by experts for taking more ethical risks because of the unexpected success of ChatGPT, partly ignoring the risks.

Chernau believes that the real threat to Google comes from open source communities, where engineers are rapidly developing artificial intelligence models that are as good as the systems of large technology companies and can be created at a lower cost. These models can be faster, more adaptable, and more useful than Google’s, Chernau says.

“We don’t have a secret recipe. The best thing for us is to learn and collaborate with what others outside of Google are doing,” he comments, predicting that customers will not want to pay for a company’s artificial intelligence models when high-quality systems are offered for free..

A Google spokesperson declined to comment on the post. At the company’s recent earnings report, Alphabet CEO Sundar Pichai painted a different picture: “Our investments and advances in AI over the past decade have put us in a good position,” pointing to advances in model development and collaboration with developers and other partners. Pichai called for the regulation of artificial intelligence, warning that the technology could become “very harmful” if it did not.

For February a large language model created by Meta leaked, spurring AI development in open source forums.. The LLaMA model is smaller than the models advertised by Google and OpenAI and is therefore easier for developers to work with. Legal researchers currently must contact Meta to access LLaMA.

“Google would do well to turn its attention to smaller and more flexible models,” Chernau said. “Giant models slow us down,” says the software engineer. “In the long run, the best models are those that can be replicated quickly.”

Source: Bloomberg.