​Google, which has developed several technologies for artificial intelligence over the past decade, and researchers have published numerous papers in the field, risks losing its edge over the open source community, warns a company engineer. Google has released many important AI technologies in recent years, but OpenAI with ChatGPT was the company that made the chatbot available to the general public.

Bloomberg detailed a message published in Google’s internal system by an engineer named Luke Sernau, the text of which was published in early April and widely discussed within the company.

Sernau’s main point is that Google, being the first to develop AI models and make its findings public, risks losing the advantage it has built up over the years to the open-source community, where many small companies are working at high speed and growing. all with unprecedented speed.

The author says that in the open source community, engineers are developing artificial intelligence models that are as good as models developed by tech giants. The difference is that these models, developed by the open source community, are cheaper to build and can be made available to customers much faster, so they pose a real threat.

Google announced a chatbot called Bard in February, but it was the most cautious launch in the company’s history, clearly under pressure from the success of ChatGPT. Google was going to release Bard, but not anytime soon. Over the past year, several top AI engineers have left Google for smaller companies that wanted to develop different AI models and systems much faster.

Sernau says Google also needs to collaborate with researchers outside the company and move to a smaller, more “nimble” model to keep up with these rapid developments.

In 2017, Google launched an important element that will underpin several AI technologies: the Transformers architecture. We are talking about natural language models, which five years ago were mainly used to solve natural language processing problems, such as machine translation, text generation or answering questions.

Google has been developing a chatbot for over two years now, but they have NOT released it to the general public because the technology is not yet complete, it would be too risky and reduce ad revenue.

In 2022, the head of Google said that launching a chatbot was a “reputational risk” for the company, but in 2021 Google announced LaMDA, a natural language processing technology that has the ability to better understand the intent and context of conversations between people.

Photo source: Dreamstime.com