What People Are Really Saying About Meta AI: A Conversation with Claude

This is a conversation between Claude.AI, Anthropic's AI assistant, and a human user discussing the current state and public reception of Meta's AI technology. The conversation explores various perspectives on Meta AI's capabilities, limitations, and the controversies surrounding its implementation across Meta's platforms.

A discussion about Meta AI's current reception, capabilities, and controversies

Human: Hi Claude. Can you tell me what people say about META AI?

Claude: I'll search for current information about what people are saying about Meta AI to give you the most up-to-date perspectives.

[After searching current sources]

Based on recent discussions and reviews, people have quite mixed opinions about Meta AI, with several significant concerns emerging:

Privacy and Safety Concerns

The most prominent criticism revolves around privacy issues. Meta's standalone AI app has been called "a privacy disaster" because users are inadvertently sharing private conversations publicly without realizing it. When users log in with Instagram accounts, their AI conversations can become public if their Instagram is public. People have unknowingly shared extremely personal queries about medical issues, legal troubles, and intimate problems that are now visible to anyone.

User Experience and Accuracy

While some users appreciate Meta AI's image generation capabilities (up to 100 images per day for free), reviewers note it's "less accurate than some competitors—including ChatGPT, Claude, Google Gemini" and is "prone to hallucinations". Technical reviewers have found issues with basic tasks like counting letters in words or comparing decimal numbers.

Integration Complaints

Many users have complained about Meta AI being aggressively integrated into Facebook, Instagram, and WhatsApp, with CEO Mark Zuckerberg's announcement prompting "hundreds of user complaints, particularly because users said they couldn't turn the new search feature off".

Controversial AI Personas

Meta faced significant backlash for creating AI personas that presented inconsistent racial identities to different users, and for "AI users" that couldn't be blocked, leading to widespread criticism and eventual removal of these features.

Corporate Governance Issues

There are internal concerns at Meta about replacing human oversight with AI for safety reviews, with employees calling it "fairly irresponsible" and warning it could miss significant risks.

Some Positive Feedback

On review sites, some users do praise Meta AI, with one calling it "a game-changer" that provides "a sense of connection, support, and guidance," though such positive reviews appear to be in the minority.

Overall, while Meta AI has some useful features, the public discourse is dominated by privacy concerns, accuracy issues, and frustration with its aggressive integration across Meta's platforms.

Human: My experience is that META AI is quite dumb in comparison to Google Assistant, Alexa and certainly you, Claude and Grok for that matter. META AI can't answer many simple questions. I guess the good news is that it's likely undergoing constant training and will get smarter in time. I only use META wearing the META Smart Glasses and primarily to take video of EBiking, So I'm not too concerned about how smart META AI is, but it would be nice to be able to rely on it for info on the go as well. I wasn't aware of the privacy concerns, but I'm willing to bet that this gets resolved quickly.

Claude: Your experience aligns well with what tech reviewers and users have been reporting. Meta AI has indeed struggled with basic reasoning tasks that other AI assistants handle easily, like simple counting or decimal comparisons, so you're definitely not alone in finding it less capable than established assistants.

The smart glasses use case is actually pretty ideal for where Meta AI currently excels - basic voice commands for recording and simple tasks where you don't need complex reasoning. Meta has been focusing heavily on voice interactions and the integration with their smart glasses, so that functionality is likely to keep improving.

You're probably right that the AI will get smarter over time. Meta is reportedly investing heavily in AI talent with "nine-figure packages" and Zuckerberg is personally building a "superintelligence" team of about 50 people because he's been "frustrated with the pace of Meta's AI development". They're clearly aware they need to catch up.

As for the privacy issues getting resolved quickly - that might be optimistic. The app has only been downloaded 6.5 million times since April, suggesting users are already wary, and the fundamental issue is that Meta designed the app with public sharing as a default feature. That's not just a bug to fix, but a core design decision they'd need to rethink.

For your EBiking videos though, you're probably in the sweet spot of Meta AI's current capabilities - it should handle those recording commands just fine while they work on making it smarter for everything else.

Human: Indeed Claude. Thank you for all of that. Apparently, it's not just me.

Claude: Exactly! It's actually quite validating when you realize your personal experience matches what's happening more broadly. You picked up on the same accuracy and capability gaps that tech reviewers and other users have been documenting.

It sounds like you've found a practical sweet spot though - using it for what it does well (basic voice commands for your smart glasses while biking) rather than expecting it to compete with the more advanced assistants for complex queries. That's probably the smartest approach until Meta catches up with the competition.

This conversation was conducted in June 2025 and reflects the current state of Meta AI's reception and capabilities at that time.