Overview of What Should An AI Device Look Like? — With Alex Himel
In this episode, Alex Kantrowitz talks with Alex Himel, Meta’s VP of Wearables, about the future of AI devices and why Meta believes glasses are the most promising AI form factor. The conversation centers on how Meta is turning Ray-Ban and Oakley smart glasses into always-on AI companions that can see what you see, hear what you hear, and help you throughout the day without pulling you out of the real world.
Himel argues that the winning AI device won’t be just about raw AI capability — it will need to be comfortable, stylish, useful without AI, and natural to wear. He also discusses Meta’s product strategy, the role of on-device AI, privacy concerns, and what kinds of tasks these glasses may eventually take over.
Key Takeaways
- Glasses are the strongest AI wearable candidate because people already wear them, they’re socially normal, and they remain useful even if the AI is off.
- Meta sees AI glasses as a way to make users more present, not more distracted.
- The company believes the best AI assistant will combine:
- a strong device/form factor
- high-quality software and AI
- a product people actually want to wear
- Meta’s long-term vision is not necessarily to replace phones, but to let wearables handle more everyday tasks that are awkward or cumbersome on a phone.
- Privacy and social comfort remain major design constraints, especially for features like facial recognition.
How Meta Thinks About AI Glasses
Why Glasses, Not Another Device?
Himel’s core argument is simple: if a device is already something people would wear anyway, then AI becomes an added benefit rather than a burden.
He compares glasses to an escalator that becomes stairs when it stops working — they still have utility even without AI. That makes them easier to adopt than a device that only exists because of AI.
He also notes that:
- headphones can handle audio use cases,
- smart speakers can help at home,
- laptops can handle work tasks,
but glasses can sit across all of those contexts while staying with you all day.
What the Future Looks Like
Himel imagines a day where the glasses act like a quiet companion:
- running with them and getting stats hands-free
- asking for shopping reminders while cooking
- getting answers for kids’ schedule questions while driving
- having meeting notes and follow-ups drafted automatically
- using them to recall names, details, or prior conversations in real time
The theme is proactive assistance that helps without interrupting.
Product Use Cases Discussed
Everyday Audio and Media
Meta says the most-used feature today is still audio:
- phone calls
- music
- listening while walking, running, skiing, or hanging out outdoors
Himel emphasized that the microphones and noise suppression are strong enough that the glasses can outperform many earbuds for calls.
Photos and Video
The second major use case is capturing photos and video hands-free.
Examples mentioned:
- recording kids’ performances while staying present
- filming a sunset or vacation moment without reaching for a phone
- capturing sports footage or real-world moments that would otherwise be missed
Meta is also launching AutoCapture, which can automatically detect moments worth recording.
AI Assistance
AI use cases are expanding from simple queries into more advanced help:
- “What am I looking at?”
- “Summarize this conversation”
- “Help me do this thing”
- “Remind me to follow up later”
- “Draft an email / message for me”
Himel repeatedly frames these as small conveniences that add up, but also as the beginning of something more transformative.
Ray-Ban Meta Optics and Product Design
One of the main product stories in the interview is the new Ray-Ban Meta Optics.
What’s Different About Them?
These new glasses are designed to be:
- smaller
- lighter
- more optical-friendly
- better for indoor and all-day wear
Meta says they’ve made hardware advances in:
- thinner arms
- slimmer frames
- better weight distribution
- adjustable nose pads
- more comfortable hinges
- easier prescription fitting via optical channels
Why This Matters
Earlier Meta glasses were more naturally positioned as sunglasses and outdoor devices. The new optical frames expand the product into workplaces, indoor environments, and all-day use, which broadens the AI assistant’s role significantly.
Wearables, Agents, and the Future of AI
The episode also explores how wearables fit into the broader shift from chatbots to agentic AI.
What “Personal Superintelligence” Means Here
Meta’s framing is that AI should help users become the best version of themselves while keeping them present and connected.
That means AI should:
- help at the right moment
- avoid spamming the user with notifications
- understand context
- automate tasks only when helpful
On-Device vs Cloud AI
Himel explains that Meta sees a split between:
- training, which will stay in data centers
- inference, which may increasingly move onto devices
Why move inference local?
- lower latency
- better reliability
- lower cost
- stronger privacy
But some tasks will still require cloud-scale models because local models may not yet be capable enough.
Privacy, Facial Recognition, and Social Acceptability
A major portion of the conversation focuses on the boundaries of what these glasses should do.
Name Tag / Facial Recognition
Kantrowitz raised reports about a possible facial-recognition-style feature. Himel said:
- it is not launched
- Meta is aware it is a commonly requested idea
- the company is being very cautious because of privacy and social discomfort concerns
Meta’s view is that if a feature makes other people feel uneasy, users may stop wearing the glasses altogether.
The Principle
Himel says Meta wants features that map to normal social behavior, but make it easier.
Examples:
- remembering a name you already learned
- recognizing someone from an event
- having a virtual version of a name badge in workplace settings
But not:
- silently identifying random people on the street
Meta’s Competitive Position
Kantrowitz pressed on whether the AI wearable winner will be the company with the best assistant.
Himel’s Answer
He said it’s a combination of:
- device quality
- comfort and style
- software quality
- AI capability
In other words, the assistant matters, but people still won’t wear a bad-looking or uncomfortable product, even if the AI is good.
He also suggested that users may eventually prefer one primary assistant, though some power users may use multiple tools for different tasks.
The Origin Story of AI in Meta Glasses
One of the most interesting moments in the interview is the backstory of how AI became central to the glasses roadmap.
Himel says that:
- the glasses were already in development before the GenAI boom
- Mark Zuckerberg sent him a long WhatsApp message while he was driving with his child in the back seat
- after that conversation, Meta pivoted 200 people to build AI for the glasses
That decision helped turn the glasses from a hardware experiment into a serious AI platform.
Notable Moments
On the Value of Wearables
“An escalator never breaks, it becomes stairs.”
On the Goal of the Product
The glasses should help you stay “present yet connected.”
On Social Norms
A feature is only worthwhile if people are comfortable with it and don’t find it creepy.
On the Bigger Vision
Wearables may eventually become ubiquitous, handling both familiar phone-like tasks and brand-new AI tasks.
Bottom Line
This episode presents Meta’s bet that AI glasses will become the most natural and useful AI device category. The strategy is not to invent a weird futuristic gadget, but to make a product people already want to wear and then layer increasingly capable AI on top.
The main idea is that the future AI device will win if it is:
- stylish enough to wear daily
- useful even without AI
- capable of seeing, hearing, and assisting in context
- private and socially acceptable
- backed by software good enough to earn trust
Meta’s answer to “What should an AI device look like?” is clear: a great pair of glasses.