AI recommendations lift time spent on Facebook and Instagram, Meta says

Meta's second-quarter earnings brought a notable engagement claim: AI-powered improvements to its recommendation systems drove a 5 percent increase in time spent on Facebook and 6 percent on Instagram, with video viewing time up 20 percent year over year. Increases of that size were under one percent not long ago.

The gains confirm that algorithmically suggested content from accounts users do not follow, however much users grumble about it, is working in the company's favor. For marketers, the implication is that original, engaging content is being actively rewarded by the recommendation engine, and reach is increasingly earned through interest-targeting rather than follower graphs.

Ahead of the call, the company's chief executive unveiled a push toward what he called personal superintelligence: AI assistants that understand and anticipate individual needs, delivered largely through smart glasses that see and hear what the wearer does. He argued AI glasses will become the main way people integrate such assistance into daily life.

The vision marks a shift from the company's earlier championing of open-source AI toward a more tightly controlled ecosystem, and the privacy questions are immediate: always-on glasses that can observe an entire environment arrive from a company already among the least trusted on data handling.

Investors looked past the skepticism. With its apps drawing more than 3.4 billion people a day and AI lifting both organic engagement and ad performance, the stock climbed after the report.

Elsewhere in the social media industry, practitioners report experimenting with AI agents that can generate and schedule a month of content from a single image in under an hour, prompting a running debate about how much automation a brand's voice can survive before it stops sounding like itself.