Google may be missing out on its advantage in the field of artificial intelligence (AI) due to a focus on rivalries with startups instead of open communities, according to Google engineer Luke Sernau. In a document published on the internal platform of Google in April, Sernau criticized the company’s lack of progress in open source technologies, believing that the real threat to Google competition from open communities that develop customized AI models quickly and cheaply. These communities could pose a real challenge to Google’s AI development. Sernau believes that Google’s best hope is to study and cooperate with others working outside the company.
Sernau expressed concern that customers may refuse to pay for models if they are high-quality and available for free. Google has not commented publicly on Sernau’s criticisms. Nevertheless, the CEO of Alphabet, Sundar Pichai, recently said that the company has made significant progress in AI with the help of developer partnerships. Sernau’s comments come after an increasing sensitivity about the regulation of AI, with many warning of the potential harmful effects of the technology.
Meta’s Llama model, which contributed to progress in the field of generative AI, caused some of the increased attention to these concerns. Sernau suggests that Google should shift its focus to smaller, more flexible models that are easier to work with. “Giant models slow us down,” Sernau wrote. “In the long run, the best models are those that you can quickly work on and improve them.”