Meta’s development of a new AI model, designed to enhance video and user Feed recommendations, marks what could be a significant shift in the online commerce landscape.
The new AI model could benefit advertisers by directing viewers to more relevant results, experts say. Meta’s progress unfolds amid a widespread industry effort to improve online search outcomes through the application of AI.
“Advertisers still have to be very specific about how their ads are recommended,” Ben Steele, an AI content developer and social media marketer for The Big Phone Store, told PYMNTS in an interview. “Upgrading Meta’s recommendation algorithm could mean that advertisers have to work a lot less hard to figure out for themselves who their target audience is, which would have the benefit of allowing companies with much smaller budgets to achieve the same level of hyper-specific recommendations that larger companies enjoy.”
Tom Alison, the head of Facebook, said at a Morgan Stanley tech conference that Meta has traditionally used separate AI models for content recommendations in various services such as Reels, Groups and Feed.
The company’s technological journey began with upgrading its recommendation systems from conventional CPUs to more robust graphics processing units (GPUs), aiming for improved efficiency. Recently, the emphasis has transitioned to investigating how large language models (LLMs) and generative AI could revolutionize the recommendation mechanism.
“Instead of just powering Reels, we’re working on a project to power our entire video ecosystem with this single model, and then can we add our Feed recommendation product to also be served by this model,” Alison said. “If we get this right, not only will the recommendations be kind of more engaging and more relevant, but we think the responsiveness of them can improve as well.”
Meta’s new recommendation algorithm is intended to work in the same way across all of their products.
“It’s not yet clear what the extent of this would be, but it’s possible that they could launch a product recommendation system similar to how ChatGPT can work,” Steele said. “LLMs are capable of recommending products, according to web searches or training data, that are hyper-specific to the user’s query.”
Alison explained that Meta has gathered a lot of GPUs to help with its AI projects, including making digital helpers. Meta plans to add better chat features to its main page, so if you see something about Taylor Swift you’re interested in, you can ask Meta’s AI for more details. They’re also trying out a feature in Facebook groups, like for baking enthusiasts, where you can request AI questions about recipes and get answers, making it easier to obtain information and interact.
The enhanced AI recommendation technologies might provide users with content that better matches their preferences. Star Kashman, a cybersecurity and privacy law expert and attorney, noted that AI can analyze vast datasets with remarkable accuracy.
“Viewers would be exposed to content that aligns more closely with their interests and behaviors, increasing user engagement and conversion rates,” she said.
However, Kashman pointed out that the new technology comes with potential downsides. She said AI could make online shopping and social media “more addictive and dangerous.”
AI is key in generating customized content, improving search engine performance, and personalizing online shopping recommendations, Kashma said.
“In order to compete, businesses must stay abreast of these advances,” Kashman said. “A culture of innovation must be fostered through investment in AI technologies, training, regulation and education. In this way, companies can refine their marketing strategies and revolutionize how they interact with their customers. However, they must utilize AI ethically and transparently.”
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More