Why Meta Might Create the Best Search Engine Ever

Meta is reportedly going after Google Search, Perplexity AI and OpenAI’s SearchGPT. Mark Zuckerberg and his team are developing a search engine that uses the web to provide conversational answers about current events via the Meta AI chatbot, which will be accessible on WhatsApp, Instagram, and Facebook.

This development follows Meta’s partnership with Reuters, enabling the chatbot to use content from the media house to answer user questions in real time about news and current events. Currently, Meta relies on Google and Bing to deliver information to users regarding news, stocks, and sports on Facebook and Instagram.

Meta Search, for Real?

Developing a search engine wouldn’t be difficult for Meta, considering its advanced search tools on Instagram and Facebook, which allow users to find each other easily on these social media platforms.

In a recent interview with Zuckerberg, Aditya Agarwal, a partner at South Park Commons, who interned at Facebook in 2005, revealed that Zuckerberg had asked him to build a search engine for the platform. “I think when people come onto Facebook, the first thing they want to do is search for people,” Zuckerberg had told Agarwal.

When Agarwal initially said he didn’t know how to build one, Zuckerberg replied, “If I can build Facebook, you can build a search engine.”

Fast-forward to today, Meta’s internal search feature ranks among the best, alongside its recommendation engine, which boosts its ability to predict user interests.

With a strong focus on research, Meta AI is making strides in projects such as MViT, XLM-R/XLM-V, and FLAVA/Omnivore to better grasp the semantic layers of content across diverse modalities—images, text, audio, and video.

Earlier this year, Zuckerberg revealed that AI now recommends over 50% of the content on Instagram and 30% of posts on Facebook. In this context, Meta’s search feature is largely driven by user activity on the platform, which informs the generation and ranking of search results. The algorithm analyses thousands of signals to determine the most relevant content for each user.

The company has created a retrieval system that, in just hundredths of a second, narrows down billions of content pieces to thousands and then to a few hundred relevant to a specific individual’s interests.

On the generative AI aspect, Meta Search is likely to be powered by Llama 4, which is set to be released by early next year and will have improved reasoning capabilities. Meta AI chief Yann LeCun refers to it as autonomous machine intelligence (AMI), which also means ‘friend’ in French. He believes these systems can truly help people in their daily lives by developing the ability to understand cause and effect and model the physical world.

Meta Boasts of Large User-base

Today, Meta AI has 500 million users worldwide, according to Zuckerberg.

“Crazy to think there’s a world where Meta could come in and win search against Google in the next 24-36 months,” said a user on X.

It makes sense for Meta to enter the search game, given the massive user base of its social media platforms. As of June 2024, WhatsApp has an estimated 3 billion unique active users worldwide, while Instagram reached over 2 billion monthly active users in early 2024.

These platforms already integrate Meta AI chat. With the new search feature, users will be able to access real-time data directly within the platforms, eliminating the need to visit Google.

The possibilities are endless. Meta recently launched Meta Ray-Ban glasses, which can answer users’ queries through voice and are multimodal, meaning they can understand and see their surroundings as well. In the future, the search capability in the glasses will provide users with real-time information simply by asking interactive questions.

Meanwhile, OpenAI recently rolled out the ability to search through chat history on ChatGPT web, allowing users to easily reference past conversations. Interestingly, according to AI insider Jimmy Apples, OpenAI was set to launch SearchGPT last week but pulled it back for one reason or another. “Not wanting to clash with Anthropic’s release?” he questioned.

Today, Google holds a user base advantage, but only time will tell how long it can maintain that lead. As of August 2024, ChatGPT has over 200 million weekly active users worldwide, while Perplexity AI has 10 million monthly active users. “Perplexity now serves over 100 million queries every week. Next stop: 100 million+ queries every day,” said the Perplexity AI chief Aravind Srinivas in a recent post on X.

In comparison, as of October 2024, Google’s estimated user base is around 4.97 billion people out of the world’s 5.44 billion internet users. Google chief Sundar Pichai, recently revealed that the company is planning to expand its generative AI search feature called ‘AI Overviews’ to more than 100 countries, making it available to more than 1 billion users.

Can Meta Really Pull it Off?

Given its history, Meta doesn’t have a strong track record when it comes to copying others’ products. For instance, last year it launched Threads as a competitor to X (formerly Twitter), but its user base is still far from reaching Instagram or WhatsApp’s status. As of August 2024, Threads crossed 200 million users.

Meta could potentially capture the advertising industry from Google through its search feature. In the recent quarter, Google Search and other advertising revenue rose by 12%, reaching $49.4 billion. Meanwhile, Instagram generated $39 billion in revenue in 2023.

Although this figure is lower, Meta still has a chance to catch up, to say the least.

Meanwhile, Perplexity AI already has plans to integrate advertising into its services. In a recent interview with a major publication, Srinivas revealed that Perplexity’s AI advertising, unlike Google’s, won’t compromise the quality of the answers.

Srinivas said they won’t tinker with answers. Instead, they would provide ads through the suggested follow-up questions. “Let’s keep the answers accurate and truthful, and let’s incentivize users to learn more about brands through suggested sponsored questions,” he said.

He explained that after each query, Perplexity suggests follow-up questions. He added that, in the future, one or two of those questions might be sponsored.

The search territory, dominated by Google for years, now appears to be losing ground, and it would be intriguing to see how the industry takes shape.

The post Why Meta Might Create the Best Search Engine Ever appeared first on AIM.

Follow us on Twitter, Facebook
0 0 votes
Article Rating
Subscribe
Notify of
guest
0 comments
Oldest
New Most Voted
Inline Feedbacks
View all comments

Latest stories

You might also like...