Stung from the hit it took from an Apple privacy feature three years ago, Meta is reportedly looking to decrease its dependence on Google and Microsoft. The Information said on Monday that Meta is developing a search engine for its chatbot. The company also recently partnered with Reuters to help its AI answer news-related questions.
Meta has reportedly been working on indexing the web for at least eight months. The company’s goal is said to be to integrate the indexes into Meta AI, giving the chatbot an alternative to Google Search and Microsoft Bing. Meta publicly disclosed its web crawler tech this summer, only saying it was for “training AI models or improving products” without stating outright that it was building a search backend. Senior engineering manager Xueyuan Su is reportedly leading the search engine project.
The Information says the move directly results from dependence on other Big Tech companies that have “stung” the company in the past, like Apple’s App Tracking Transparency (ATT). Meta previously said the iPhone privacy feature introduced in 2021 would cost over $10 billion in lost ad revenue. (It was so concerned about the feature it was sued for allegedly evading its rules.)
The thinking is that CEO Mark Zuckerberg wants Meta to be as autonomous as possible to avoid a repeat of that if Google or Microsoft were to cut off its access to web searches. It’s unknown if Meta currently pays either company for that access.
Engadget reached out to Meta for comment. We’ll update this story if we hear back.
The company’s AI investments appear to be paying off. Zuckerberg posted on Threads in August that Meta AI has over 185 million weekly active users and more than 400 million monthly. “Growing quickly, and we haven’t even rolled out in UK, Brazil, or EU yet,” the Facebook founder wrote. OpenAI said early this month that ChatGPT had 250 million weekly users.
This article originally appeared on Engadget at https://www.engadget.com/ai/meta-is-reportedly-developing-a-search-engine-for-its-chatbot-172505704.html?src=rss
