Meta (NASDAQ:META) is going all in on AI, with CEO Mark Zuckerberg pledging to spend hundreds of billions over the next few years. His vision? A personal superintelligence baked into everything from WhatsApp to Ray-Bans. But Meta’s flagship productthe standalone Meta AI app launched in Aprilisn’t quite delivering. While it boasts 1 billion monthly users thanks to deep integration across Facebook, Instagram, and Messenger, the experience feels fractured. Conversations don’t carry over between apps. The same chatbot responds differently depending on where you use it. And personalization? Still surface-level. Bloomberg testing found Meta AI could guess your zip codebut still serve up generic weekend recs pulled from Eventbrite.
The feedback loop isn’t kind either. On Reddit (NYSE:RDDT), a post titled Who hates Meta AI? drew over 72,000 upvotes, with replies calling it unnecessary and slop. The Discover feed, designed to inspire, often shows outdated or bizarre image generations, like a dog in a yellow hat or anime-style selfies. And while Meta says user chats aren’t shared publicly unless manually posted, earlier versions of the app reportedly published private conversations without users realizing it. Meanwhile, ChatGPT and Perplexity offer memory across sessions and platforms, letting users pick up where they left offsomething Meta’s assistant can’t do, even when all accounts are linked.
To be fair, Meta’s scale gives it a powerful runway. Llama, its open-source language model, is already being built into third-party tools. Meta AI is also integrated into hardware like the Quest headset and Motorola’s Razr foldables. And Zuckerberg still sees the U.S. app as the core battlefield for AI dominance. But to win that war, Meta has to fix more than a few bugs. Right now, its AI strategy feels more like a patchwork experiment than a polished product. If this is step one, investors will want to see much sharper execution in the next few.
This article first appeared on GuruFocus.
