Tell Me Why

AI Smart Glasses Are Getting Smarter — and How Metalenses Could Make Them Smaller

Imagine walking through a city where you do not speak the local language. You look at a restaurant sign, ask your glasses what it says, and hear the translation without taking a phone out of your pocket. A camera captures what you are seeing, artificial intelligence interprets the scene, and the answer reaches you through a speaker or, in some devices, appears on a small display in front of your eye.

Parts of this experience already exist. Modern smart glasses can capture first-person images, listen to voice commands, answer questions about the surrounding environment, translate conversations, and in some cases display information directly in the wearer’s field of view.

At the same time, a very different technology is advancing inside research laboratories: the metalens. Instead of using several conventional curved lenses, a metalens manipulates light with extremely small engineered structures placed across a thin surface. If the technology can be manufactured reliably at scale, it could help make some cameras, eye-tracking systems, and wearable optical components significantly smaller.

Artificial intelligence and metalenses solve different problems. AI makes glasses more capable of understanding the world. Metalenses may help engineers reduce some of the physical hardware needed to see it.

Smart Glasses Are Becoming Eyes and Ears for AI

Traditional glasses are passive optical devices. AI-powered glasses are miniature computing systems positioned close to the wearer’s eyes and ears.

Depending on the model, smart glasses can include cameras, microphones, speakers, motion sensors, wireless connectivity, processors, batteries, and sometimes an integrated display. Some processing happens directly on the device, while more demanding AI tasks may depend on a connected smartphone or cloud infrastructure.

The camera itself does not understand what it sees. It converts incoming light into digital information. Computer-vision algorithms and multimodal AI models then analyze that information and attempt to identify objects, text, scenes, and other visual details.

A simplified process looks like this:

Physical scene → optical lens → image sensor → digital image → computer vision and AI → response to the user

The response may be spoken through an open-ear speaker or displayed visually when the glasses contain a screen.

AI Glasses and AR Glasses Are Not the Same Thing

It is important to separate camera-based AI glasses from full augmented-reality glasses.

Some smart glasses use cameras, microphones, speakers, and AI assistants but do not project digital graphics across the wearer’s vision. They primarily communicate through audio.

Display glasses add a visual interface capable of showing information such as messages, captions, directions, or translations.

Full augmented-reality systems go further by attempting to place digital objects or information into a wider view of the physical environment.

This distinction matters because the engineering requirements are dramatically different. Adding a camera and an audio assistant is easier than building lightweight glasses with a bright, wide-field display, long battery life, strong computing performance, and ordinary-looking frames.

What AI Glasses Can Do Today

Meta is one of the companies pushing AI glasses into the consumer market.

Ray-Ban Meta glasses combine cameras, microphones, speakers, wireless connectivity, and Meta AI. Depending on the supported region and software features, users can take photographs, record video, interact with an AI assistant, ask questions about what they are looking at, and use translation functions.

Meta has also developed display-equipped glasses that move beyond audio-only interactions. A display can make tasks such as navigation, captions, messages, and translated text more immediate because the user does not need to look at a separate phone screen.

Google is pursuing a similar direction through Android XR and Gemini-powered eyewear. The company has demonstrated glasses designed to use cameras, microphones, speakers, AI, navigation, translation, and smartphone integration.

However, product demonstrations, announced devices, research prototypes, and widely available consumer products should not be treated as the same thing. A technology can work convincingly in a controlled demonstration long before it becomes affordable and reliable enough for millions of people.

Real-Time Translation Could Become One of the Most Useful Features

Translation is a natural application for smart glasses because the device is already positioned near the eyes and ears.

A traveler could look at a sign, menu, ticket, or warning notice and ask the AI to explain or translate it. Microphones can also capture spoken language and convert it into translated speech or captions.

Several technologies have to work together for this experience to succeed.

The system may first need speech recognition to determine what was said. It then needs a translation model to interpret the meaning. If text appears in the physical environment, optical character recognition can identify the written words before they are translated.

Display-equipped glasses could eventually place translated text directly into the user’s field of view. Audio-focused glasses can instead speak the translation through built-in speakers.

Real-time translation is not perfect. Performance can be affected by accents, background noise, poor lighting, unusual vocabulary, internet connectivity, and the languages involved. Errors that are acceptable while ordering food may be much more serious in medical, legal, or safety-critical situations.

Why Multimodal AI Changes the Experience

The most important software development may be multimodal AI.

Traditional digital assistants mostly waited for text or spoken commands. Smart glasses can potentially give an AI system visual context as well.

Instead of describing an object, the user can simply look at it and ask, “What is this?”

Instead of typing a foreign sign into a translation application, the user can ask, “What does that sign say?”

A multimodal AI system may combine images, speech, text, and contextual information to produce an answer.

This could support questions such as “What building am I looking at?”, “Read this menu,” “What does this warning mean?”, or “Which direction should I walk?”

Some of these capabilities already exist in limited forms. Others still depend on better reliability, lower latency, improved battery efficiency, and stronger integration between wearable devices, smartphones, and cloud AI.

Smart Glasses Could Also Improve Accessibility

Wearable computer vision may have particularly important applications for people with visual or hearing impairments.

A first-person camera can potentially help describe surroundings, read text aloud, identify objects, provide navigation assistance, or connect a user with remote visual assistance.

Meta has worked with Be My Eyes, a service designed to connect blind and low-vision users with visual assistance.

Smart glasses may also support live captions, environmental descriptions, translation, and hands-free access to information.

These systems should not be described as replacements for established mobility aids or professional assistance. Their value is in providing another layer of contextual information.

The Hardware Problem: Intelligence Takes Space

Software receives much of the attention, but smart glasses face severe physical constraints.

Engineers have to fit cameras, optical systems, processors, batteries, antennas, microphones, speakers, sensors, and sometimes displays into frames that people are expected to wear for hours.

Every additional component adds size, weight, power consumption, and heat.

A smartphone has enough room for a relatively large battery and cooling system. Glasses do not.

This is one reason optical miniaturization has become an important research area.

What Is a Metalens?

A conventional optical lens normally uses curved glass or plastic to bend light toward a focal point.

A metalens takes a different approach.

It uses a thin surface covered with extremely small engineered structures often described as meta-atoms. These structures can be smaller than the wavelength of the light they manipulate.

By carefully controlling their size, shape, and arrangement, engineers can change how different parts of a light wave travel through or interact with the surface.

The result can be a very thin optical element capable of focusing light.

This field is closely connected with metasurfaces, which use specially designed microscopic or nanoscopic structures to control optical properties.

A metalens does not automatically make an entire camera flat. Cameras still require an image sensor, electrical connections, packaging, processing hardware, and physical spacing between some components.

The potential advantage is that part of the conventional optical stack may be redesigned or compressed.

Why Metalenses Are Interesting for Smart Glasses

Size matters much more in wearable technology than in many other electronic devices.

Saving a few millimeters inside a desktop camera might not matter. Saving the same space near the front of a pair of glasses could make the frame thinner, lighter, or more comfortable.

Metalenses are therefore being studied for applications including miniature cameras, eye tracking, iris authentication, depth sensing, near-eye displays, augmented reality, virtual reality, medical imaging, and machine vision.

A particularly relevant research result was published in Nature Communications in 2025.

Researchers demonstrated a compact infrared eye camera based on a metalens with a field of view of approximately 120 degrees and a total optical track length of only 1.758 millimeters.

The researchers demonstrated applications including pupil tracking and iris authentication, both of which are highly relevant to future AR and XR devices.

This was a research prototype rather than a consumer smart-glasses camera, but it shows why flat optics are receiving serious attention from wearable-device researchers.

Metalenses Are Getting Closer to Image Sensors

Another major challenge is integrating metalenses with commercial image sensors.

In 2025, researchers reported achromatic metalens arrays integrated with a commercial imaging chip. The study demonstrated full-color imaging and explored manufacturing approaches compatible with semiconductor processes.

That type of integration is important because a lens that works beautifully by itself is much less useful if manufacturers cannot reliably connect it to sensors, electronics, and mass-production processes.

The long-term opportunity is not simply to replace one glass lens with a flat lens. Engineers may eventually be able to redesign the entire camera module around much thinner optical structures.

The Full-Color Problem

One of the most difficult optical problems for metalenses is chromatic aberration.

Visible light contains many wavelengths. Red, green, and blue light do not automatically behave identically when they pass through an optical system.

If different colors focus at different distances, the resulting image can suffer from blur, color errors, and reduced sharpness.

Researchers have therefore developed achromatic metalenses designed to focus multiple wavelengths more consistently.

A study published in Nature Communications demonstrated achromatic metalenses capable of operating across a broad portion of the visible spectrum and producing full-color images.

Researchers have also explored larger RGB metalenses and manufacturing techniques designed for near-eye displays.

In 2025, a Nature Materials paper demonstrated a centimeter-scale RGB achromatic metalens manufactured using a roll-to-plate printing approach for wide-field holographic near-eye displays.

Research published in 2026 also demonstrated a compact full-color near-eye display using an RGB achromatic metalens integrated with an OLED panel.

These experiments show how the field is moving beyond the simple question of whether a flat lens can focus light. The harder questions now involve color accuracy, efficiency, field of view, manufacturability, and integration with real electronic systems.

Conventional Lenses vs. Metalenses

Factor	Conventional Optics	Metalenses
Light control	Curved refractive elements	Nanostructured metasurface
Physical form	Often several curved elements	Potentially very thin and flat
Manufacturing maturity	Extremely mature	Still developing rapidly
Full-color performance	Highly developed	Improving but technically challenging
Mass production	Well established	Still a major challenge
Sensor integration	Mature commercial ecosystem	Promising research demonstrations
Wearable potential	Already widely used	Potentially valuable for miniaturization

Manufacturing May Be the Biggest Challenge

Producing one advanced metalens in a research laboratory is very different from producing millions of identical lenses for consumer electronics.

The nanostructures in a metalens require extremely precise fabrication. Small variations in their dimensions or position can affect how light is manipulated.

Manufacturers must also consider defects, production yield, material selection, alignment with image sensors, wafer-scale manufacturing, quality control, and cost.

Researchers are investigating semiconductor-compatible fabrication, nanoimprint lithography, high-aspect-ratio nanostructures, and scalable printing methods.

A 2026 review published in Nature Reviews Electrical Engineering highlighted continuing trade-offs involving focusing efficiency, numerical aperture, spectral bandwidth, field of view, device size, and manufacturability.

This makes the current position of metalenses more nuanced than the headlines sometimes suggest.

They are no longer merely theoretical optical devices, but they are not yet universal replacements for conventional camera lenses.

Could Metalenses Make Smart Glasses Look Like Ordinary Glasses?

Possibly, but only as part of a much larger engineering effort.

Reducing the thickness of a camera or eye-tracking system could save valuable space. However, smart glasses still require batteries, processors, microphones, speakers, antennas, wireless hardware, sensors, displays, structural components, and thermal management.

True AR glasses introduce even more difficult optical requirements because they need to generate visible digital information and deliver it into the user’s eyes while maintaining acceptable brightness, image quality, weight, and power consumption.

A metalens could help shrink some optical components without solving every other problem inside the frame.

The path toward glasses that look almost indistinguishable from ordinary eyewear will probably require progress across batteries, chips, displays, optics, materials, and AI simultaneously.

The Privacy Problem

Smart glasses introduce privacy questions that are different from those created by smartphones.

A phone camera normally has to be visibly lifted and pointed at a subject. A glasses camera naturally faces approximately wherever the wearer is looking.

That raises questions about bystander privacy, audio recording, facial recognition, biometric information, location data, cloud processing, and long-term storage.

Some manufacturers use visible indicators to signal when recording is taking place. Other protections can include on-device processing, encryption, data minimization, automatic deletion, application permissions, and restrictions on biometric identification.

Technical protections alone may not be enough. Social acceptance will also influence whether people are comfortable around wearable cameras.

Cybersecurity Matters When the Computer Is on Your Face

AI glasses combine cameras, microphones, wireless communication, user accounts, cloud services, and potentially biometric information.

That creates a meaningful cybersecurity challenge.

Threats could include compromised user accounts, malicious applications, insecure cloud storage, unauthorized camera or microphone access, firmware vulnerabilities, wireless attacks, and leakage of sensor data.

Future glasses may also use eye tracking, iris authentication, or other biometric systems. Protecting biometric information is especially important because unlike a password, a person's physical characteristics cannot simply be replaced after a data breach.

Security therefore needs to be designed into the hardware, software, network connection, and cloud infrastructure from the beginning.

AI Glasses May Complement Smartphones Before Replacing Them

Smart glasses have one major advantage over phones: position.

They are already close to the eyes and ears, their cameras naturally follow the wearer’s direction of attention, and they can provide information without requiring the user to continuously hold a device.

This makes them attractive for translation, navigation, contextual AI, photography, accessibility, and short interactions.

Smartphones still have major advantages of their own.

They offer large displays, much larger batteries, powerful processors, mature application ecosystems, easier typing, better media consumption, and more space for cooling and hardware.

The most realistic near-term future may therefore involve cooperation rather than replacement.

Glasses could handle fast contextual interactions while the smartphone continues to perform tasks that benefit from a large screen or heavier computing.

What Is Available Now, What Is Emerging, and What Is Experimental?

Available Now

Commercial smart glasses can already capture first-person photos and video, provide voice-based AI assistance, perform some visual analysis, and support translation in selected languages and regions. Display-equipped models can also show information such as messages, captions, navigation, and translated text.

Emerging

AI eyewear platforms are expanding, while companies are working on deeper integration between multimodal AI, smartphones, displays, and wearable hardware. Metalens research is also advancing toward full-color imaging, sensor integration, wider fields of view, and manufacturing methods designed to scale.

Experimental

Lightweight glasses that combine wide-field augmented reality, powerful continuous AI, advanced computer vision, long battery life, and the appearance of ordinary eyewear remain extremely difficult to engineer.

Many of the most impressive metalens systems are also still research prototypes rather than components inside mass-market consumer glasses.

The Real Story Is Convergence

The future of smart glasses is unlikely to depend on a single breakthrough.

It depends on many technologies improving together.

AI models need to become more capable and efficient. Processors must consume less power. Batteries need to provide more energy without making frames heavy. Displays need to become brighter and smaller. Cameras and sensors need to shrink. Wireless connectivity needs to improve. Optical systems need to become more compact.

Metalenses fit into this larger technological convergence because they may reduce the size of some optical systems.

AI provides another part of the equation by turning raw sensor information into useful context.

A camera without intelligence simply collects images. An AI assistant without access to the surrounding environment lacks visual context.

When compact sensors and powerful multimodal AI are combined in wearable hardware, the interaction becomes much more natural.

A traveler could look at a menu and hear a translation. A person with low vision could request a description of an object. A pedestrian could receive navigation information without staring at a phone. A technician could eventually access instructions while keeping both hands free.

The future may not look like science-fiction holograms covering everything we see.

The more important change could be much quieter: glasses that understand enough about the surrounding world to provide useful information exactly when it is needed.

My Comment

The most significant development is not AI glasses by themselves and it is not the metalens by itself. It is the convergence of increasingly capable software with optical hardware that can become smaller, lighter, and more tightly integrated with sensors. Metalenses could eventually remove valuable millimeters from imaging, eye-tracking, and display systems, while multimodal AI could make those sensors far more useful. But miniaturization alone will not determine whether smart glasses become a major computing platform. A technically impressive device can still fail if it is uncomfortable, has weak battery life, makes frequent AI errors, creates privacy concerns, or simply looks like something people do not want to wear. Reliability, comfort, privacy, security, and social acceptance may ultimately matter just as much as the next optical breakthrough.

Sources

Meta: Ray-Ban Meta and AI glasses information
https://www.meta.com/ai-glasses/

Meta Newsroom: Meta Ray-Ban Display and AI glasses
https://about.fb.com/news/2025/09/meta-ray-ban-display-ai-glasses-emg-wristband/

Google: Android XR and Gemini-powered smart glasses
https://blog.google/products-and-platforms/platforms/android/android-xr-io-2026/

Be My Eyes: Accessibility and smart-glasses integration
https://www.bemyeyes.com/

Nature Communications: Compact metalens-based eye camera
https://www.nature.com/articles/s41467-025-62577-1

Nature Communications: On-chip integration of achromatic metalens arrays
https://www.nature.com/articles/s41467-025-62539-7

Nature Materials: RGB achromatic metalens for near-eye displays
https://www.nature.com/articles/s41563-025-02121-0

Nature Communications: Achromatic metalens for full-color near-eye displays
https://www.nature.com/articles/s41467-026-73103-2

Nature Reviews Electrical Engineering: Challenges and opportunities of metalenses
https://www.nature.com/articles/s44287-026-00276-9