Meta claims it used artificial intelligence to boost traffic to its Facebook and Instagram platforms and improve the earning potential of ads. Speaking on the company’s earnings call on Wednesday, CEO Mark Zuckerberg said the company was “no longer behind in building AI infrastructure”. It comes off the back of a major pivot for the Facebook parent company away from building the metaverse to focusing on commercialising its AI research.
The launch of OpenAI’s natural language AI platform ChatGPT in November marked a step change in the tech sector. Companies from Microsoft and Google to Salesforce began focusing on generative and foundation model AI and ways to integrate the technology.
Meta had already invested heavily in artificial intelligence research, often as part of its metaverse push. It published the LLaMA large-language model, created a foundation AI model to verify Wikipedia citations, published an AI chatbot months before ChatGPT and released image, 3D and video tools.
However, much of this was released or designed for research purposes, or part of the wider metaverse play, rather than as something it could capitalise on commercially. Zuckerberg says the company now has “the capacity to do leading work in this space at scale”.
While slow to adopt to the commercialisation of AI, Meta carried out a major overhaul of its core business including improving AI capacity through hardware and software systems. It says this has helped reduce overall costs and improve performance, leading to it beating first-quarter profit and revenue expectations.
The biggest problem facing Meta was one of scale. While it had the software, the research and the underlying technology, its systems were not geared towards the power-hungry compute demands of artificial intelligence models. There were also gaps in tooling, workflows and processors that required heavy investment to solve. This included changing physical infrastructure design, platform stability and software systems, an investigation by Reuters discovered.
Meta engineers spent months improving systems, including upgrading data infrastructure, relabeling information and allowing for exabyte-scale processing. “Over time, the data platform has morphed into various forms as the needs of the company have grown,” Meta’s engineers wrote in a blog post published in January detailing the data migration.
“What was a modest data platform in the early days has grown into an exabyte-scale platform. Some systems serving a smaller scale began showing signs of being insufficient for the increased demands that were placed on them.”
Meta’s platform update and migration
The efforts appear to have paid off with Meta confirming AI recommendations increased time spent on Instagram by 24% in the last quarter compared to the same time the year before. “Our investment in recommendations and ranking systems has driven a lot of the results that we’re seeing today across our discovery engine, Reels, and ads,” Zuckerberg told investors.
“Along with surfacing content from friends and family, now more than 20% of content in your Facebook and Instagram feeds i recommended by AI from people, groups, or accounts that you don’t follow,” the CEO explained, outlining wider uses of AI at a product level.
As well as increasing use of AI for recommendations, Meta is also focusing on foundation models such as those behind Midjourney and ChatGPT. The aim, said Zuckerberg, is to find new use cases including generative AI. “The work happening now is going to impact every single one of our apps and services,” he said. “I’m incredibly excited to ship more of the things that we’re building over the coming months.”
The company is expected to bring AI agents into its chat applications Messenger and WhatsApp, as well as image and video generation tools into Facebook, Instagram and the ads platform. “I expect that these tools will be valuable for everyone from regular people to creators to businesses. For example, I expect that a lot of interest in AI agents for business messaging and customer support will come once we nail that experience. Over time, this will extend to our work on the metaverse too, where people will much more easily be able to create avatars, objects, worlds, and code to tie all of them together.”