Google announces "red code": the company is looking for the answer ChatGPT

Google announced a " red code " - as a reaction of the company's bosses to the appearance of ChatGPT. The New York Times writes about this, citing an internal company memorandum and audio recordings of several meetings that were available to the publication.

Just last week at a general staff meeting, both Sundar Pichai and Jeff Dean convinced colleagues that ChatGPT could not threaten Google's search and business model; that the company has its serious developments in this area, but haste and the withdrawal of a raw product will be disastrous for Google; and that by 2023 they will have their own AI solutions ready. And within a few days, the company managed to hold a series of meetings with the participation of the CEO, where they tried to determine the strategy for the development of AI products. And according to journalists, “numerous divisions” of Google have now been transferred to the AI ​​direction.

Both the company's memorandum and audio recordings of the meetings confirm that Google sees OpenAI 's ChatGPT as a threat to its search engine and is looking for a symmetrical response.

At the same time, the AI ​​direction developed in the corporation and to all the "red codes". Almost everyone wrote about LaMDA, a language model for conversational applications, this summer after Google engineer Blake LeMoine said the company's chatbot was intelligent.

There is no doubt that Google will be able to use its developments in search engine optimization. The problem, according to experts, lies in the fact that the perfect chatbot search engine can "cannibalize" the company's business model. “If Google gives you the perfect answer to a query, you won’t click on a single advertising link,” said Amr Awadallah, who has worked at both Google and Yahoo, and now runs Vectara, a startup that is also trying to use neural networks for search.

At the same time, Google is working on a competitor to DALL-E, which could create images no worse than another well-known OpenAI development.

ChatGPT, which has amassed more than 1 million users in just five days of launch, is able to generate unique responses to queries in the style of live human speech, synthesizing information from millions of sites.

But some quickly noticed that the bot was not yet perfect. Insider talked to experts, and they were unanimous: ChatGPT cannot guarantee the complete accuracy of information and does not always distinguish between a verified fact and misinformation. From time to time, the bot begins to “hallucinate” and come up with answers that are completely out of touch with reality.

The annual Google I/O developer conference will take place in May 2023. Many suggest that it can become a platform for presenting the company's AI products.