Google engineer concerned about open-source AI gaining edge

A leaked document from a senior software engineer sounds the alarm about Google’s artificial intelligence development.

According to a critique written by a senior software engineer at Google, the company’s artificial intelligence (AI) development is threatened by the open-source community. The engineer contends that independent researchers in open-source technology are making swift and unforeseen strides in AI technology. 

In early April, Luke Sernau, a software engineer at Google, published a document on an internal system. The document was subsequently shared extensively among Google employees over the next few weeks. A source, who requested anonymity as they were not permitted to discuss internal company affairs, claimed that the document had been circulated thousands of times. On Thursday, the document was published by consulting firm SemiAnalysis and began to circulate in Silicon Valley.

According to Sernau’s assessment, Google’s competition with OpenAI had diverted attention from the swift progress in open-source technology. In his document, Sernau wrote that Google had focused excessively on keeping tabs on OpenAI’s progress, and while both companies struggled to outdo each other, open-source technology had quietly advanced. Sernau claimed that this third group, open source, was surpassing Google and OpenAI in the AI race.

Google is known for investing in futuristic technologies, and its labs have played a significant role in developing the current AI-powered chatbots. However, OpenAI, a startup, has emerged as a frontrunner in generative AI, which involves software that can create its own images, text, and videos. OpenAI’s ChatGPT was launched in November and quickly gained popularity. Its sudden success has left Google sprinting to catch up in a key subfield of the technology.

Related: Google DeepMind CEO Demis Hassabis says we may have AGI ‘in the next few years’

However, Sernau argued that the open-source communities pose a real threat to Google, as engineers are developing models that can rival the quality of those at big tech companies at a faster pace and lower cost. According to him, these models are more customizable, faster and more useful than Google’s own models. He also worried that clients may not be willing to pay for high-quality technology when it is available for free in open-source communities.

Magazine: How to control the AIs and incentivize the humans with crypto