Categories
Advice and Guidance

Smart Glasses in Education

Photo by Panos Sakalakis on Unsplash

Why look at smart glasses now?

Smart glasses have been commercially available in some form for over a decade. More recently, they have returned to popular conversation, driven largely by Meta and its growing range of glasses. The lack of widespread adoption has been attributed to the high cost and limited capabilities of earlier models, many will be reminded of the hype and disappointment surrounding Google Glass. Newer commercial devices are increasingly more affordable, in the hundreds rather than thousands of pounds, and claim to have considerable capabilities.

Smart glasses remain a divisive technology though. Some models include cameras while looking little different from a standard pair of glasses, bringing privacy concerns to the forefront. There is still limited research into public perceptions of smart glasses but a 2024 study of over 1,000 Australian residents found a clear divide between owners and non-owners of glasses. Non-owners expressed strong concerns around privacy and the potential for inappropriate use, fears that are unfortunately being realised – “I was secretly filmed with smart glasses and then trolled online”.

Their re-emergence raises both curiosity and concern within education.

  • How should their use by students and learners be managed?
  • And do they present any meaningful opportunities for teaching and learning?

To explore the relevance of smart glasses for education today, we will be drawing on three overlapping perspectives that have emerged from Jisc’s community groups: Artificial Intelligence (AI), Extended Reality (XR) and Assistive Technology (AT)

AI underpins many core features of the glasses, from speech recognition and image description to the integration of generative AI models that significantly expand what these devices can do. Smart glasses also have a history as AT, with products designed for captioning or visual support available for several years, albeit with limited uptake. Extended reality comes into play where smart glasses incorporate heads-up displays or augmented reality (AR). Smart glasses lightweight design can make accessing XR easier than virtual reality headsets, which can only be worn for short periods of time.

Part of a series, this first piece begins by examining what is currently being sold as smart glasses, comparing a range of available models to establish a clearer picture of their capabilities. In our next blogs, we will explore potential opportunities for teaching and learning, followed by a closer look at the key risks and barriers to their use in education.

We explore glasses against a series of aspects, you can use the links below to jump to different sections:

 

What do we actually mean by “smart” glasses?

Like smartphones, the term “smart glasses” covers a wide range of capabilities rather than a single, agreed definition. Depending on the product, “smart” might refer to internet connectivity, Bluetooth pairing, AI assistants, integrated cameras and microphones, sensors, heads-up displays, or some combination of these.

There is no standard that defines what features smart glasses must include, and companies are free to apply the label as they see fit. As a result, products marketed as smart glasses vary widely in function, design and cost. Prices range from tens of pounds to several thousand. Some glasses rely heavily on a paired smartphone for processing, while others operate as standalone devices. Some have augmented reality capability, overlaying digital information into the wearer’s field of view, while others deliver information primarily through audio or via a companion app on a phone.

These differences matter. If we are considering how smart glasses might be used, managed or restricted in education, it is essential to understand just how broad this category has become.

In this post, we compare a selection of smart glasses that are available now, or expected to be available in the UK within the next few months. The products include both assistive technology designed for disabled users and more general consumer devices. The aim is not to recommend specific products, but to illustrate the range of capabilities across devices.

 

What’s currently on offer?

For this comparison, we looked more closely at a small group of products that span different price points and intended uses:

Price and processing trade-offs

Model Full Retail Price Smartphone required for most features?
Envision Glasses £2,158.80–£3,598.80 (based on edition) No
Envision Ally Solos Glasses £517+ Yes
Ray-ban Meta Gen 1 £299+ Yes
Ray-ban Meta Gen 2 £379+ Yes
Meta Ray-Ban Display $799 (Awaiting UK price) Yes
RayNeo Air 3s Pro smart AR glasses £299.00 Any media device
XRAI ar2 Captioning Glasses £699 Yes

Table 1: Comparing full retail pricing and whether smartphones are required for the majority of features.

Smart glasses have existed for years, but widespread interest has historically been limited by the prohibitive cost. Many earlier devices were priced in the thousands of pounds, particularly those designed as assistive technology. More recently, both general consumer and assistive models have appeared at much lower price points, even in the low hundreds of pounds.

One reason for this shift is the method of offloading processing to the user’s smartphone. Instead of building powerful processors into the glasses themselves, some products rely on a paired phone with a companion app to handle AI tasks and data processing. This reduces hardware costs, but introduces trade-offs – the user needs a suitable smart phone, latency may increase, and performance will depend on the quality of the connection to the glasses.

Envision, for example, notes that for users of their Ally Solos glasses, tasks routed through the smartphone app may take longer to complete and voice commands longer to process. They advise users to pause slightly when issuing commands or waiting for image descriptions. Standalone devices avoid this issue but as we see are generally considerably more expensive, like the standard Envision Glasses model.

Sometimes companion apps can introduce additional costs, though these are often optional subscriptions for additional features. The Ally Solos glasses come with 1 year of Ally Pro subscription, valued at $200. Users not continuing that subscription can continue with the free version, but there are some feature limitations and a cap of 10 minutes for each conversation with the AI assistant. Similarly, XRAI’s captioning glasses comes with 3,600 ‘pro minutes’ which provide access to better transcription models than the free service, once used up users can choose to purchase more.

Glasses built as assistive technology may look to be on the higher price end when compared to newer consumer products like Meta’s Ray-Bans (Gen’s 1 and 2). There is considerable interest, therefore in whether commercial glasses can be affordable and effective assistive tools. We see though that products like the Ally Solos glasses look to offer a balance, combining assistive design with smartphone-based processing to reduce price.

Comparisons should also consider the specific design work, and specialist support that products built as assistive technology may provide. Tools like XRAI’s captioning glasses and Envision’s models are designed to deliver functionality to meet specific use cases rather than attract broad consumer appeal.

 

Data privacy and AI training

Model Is user data use in AI training possible?
Envision Glasses No
Envision Ally Solos Glasses No
Ray-Ban Meta Gen 1 Yes
Ray-Ban Meta Gen 2 Yes
Meta Ray-Ban Display Yes
RayNeo Air 3s Pro smart AR glasses No
XRAI ar2 Captioning Glasses No*

* if using Pro minutes some third party processors may need checking individual Ts and Cs

Table 2: Identifying whether there are clauses in the terms and conditions that allow the use of the user’s data for training AI models.

As with other tools, data handling varies significantly between providers, and the differences are not always obvious. A common concern with AI-powered tools is whether user data might be used to train an AI model. These varied considerably and demonstrated that it remains essential to check the individual terms and conditions, and privacy policies.

For the assistive technology products we looked at, data practices were relatively straightforward. Envision states that user data is stored only to provide backups and is not used to train AI models. XRAI processes captions locally on the user’s phone and deletes recordings automatically unless the user chooses to save them. When users opt for cloud-enhanced transcription using pro minutes, they can select third-party providers such as AWS, Deepgram or Microsoft Azure, and are advised to check those providers’ data practices separately.

Meta’s glasses present a more complex picture. Turning on the AI features opens multiple routes for Meta to store and process your information, with different rules depending on the category of data collected. Meta distinguishes between ‘Additional data’ (usage-related), ‘Voice Activity’ (anything said after “Hey Meta”), ‘Media’ (audio, video and photos), and data processed temporarily for live features like translation and livestreaming.

Because most stored data passes through the companion Meta app, it falls under Meta’s main Privacy Policy. According to which, ‘Additional data’ is only used if you opt in, otherwise it cannot be used for product-improvement or AI development.

Both ‘Media’ and ‘Voice Activity’ can be used for AI training, and opting out of this is not as straightforward as clicking a button. All people (with and without Meta accounts) can exercise a Right to Object to their data being used for “AI at Meta”. For UK users this is generally respected under UK GDPR, but only applies to data processed after the objection. Users objecting will need to send their objection and wait for a response from Meta to confirm.

Users can also choose to manually delete their Voice Activity to prevent this being used. The live translation feature does delete audio when the session ends, and livestreams aren’t stored on the glasses or in the app, though viewers or the user themselves could still technically record streams separately. Meta also states that they may share any information you send to Meta AI with third parties under its AI Terms.

As with generative AI tools more broadly, the key is that users need to check terms and conditions carefully and understand what they are agreeing to.

 

Recording capabilities & live streaming

Model Recording Capability Type(s) of media
Envision Glasses Yes Photo
Envision Ally Solos Glasses Yes Photo
Ray-Ban Meta Gen 1 Yes Photo, audio, video
Ray-Ban Meta Gen 2 Yes Photo, audio, video
Meta Ray-Ban Display Yes Photo, audio, video
RayNeo Air 3s Pro smart AR glasses No N/A
XRAI ar2 Captioning Glasses Yes Audio, transcripts

Table 3: Comparing the types of media the glasses can record

Recording is one of the most sensitive aspects of smart glasses use, particularly where it is not obvious when recording is taking place. Of the products we considered, only Meta’s models support both audio and video recording. Even then, recordings are limited to three minutes per clip, although users can work around this by stitching clips together or using livestreaming features (there is plenty of advice online on how to do this).

Video recording limitations certainly look to be a common drawback of smart glasses. In searching for other models with longer recording time, we did find others claiming recordings from 10 minutes up to 5 hours (e.g., BleeqUp). Those with the longest times looked to be built with recording capability at the forefront and marketed more as ‘camera glasses’ than smart glasses, aimed at capturing events like extreme sports rather than supporting daily tasks.

Envision glasses can be used to capture photographs. This functionality is essential to provide the user with audio descriptions of the world around them, but also can make it easier for blind and low-vision users to take photographs to keep and share. XRAI’s captioning glasses can be set to save audio recordings and a transcript of the captions they have created.

Privacy indicators vary. Envision’s glasses have a visible  LED indicator when photos are taken. Meta’s glasses also include an LED light that activates during photo and video recording and prevents capture from taking place if it is covered. However, a recent BBC article exploring cases of women who were unknowingly filmed with Meta smart glasses, found that there were workarounds available to disable the light and still record.

As with other recording equipment it is the user’s responsibility to ensure anything that they record is done legally. Notably, several glasses terms included reminders to the users not to record personal information with them or use them in potentially sensitive locations like doctor’s offices, places of worship and public toilets.

 

AI-powered features

Model Generative AI Functionality? AI Model(s)
Envision Glasses Yes Chooses model based on task (incl. Llama, OpenAI GPT models, Gemini)
Envision Ally Solos Glasses Yes Chooses model based on task (incl. Llama, OpenAI GPT models, Gemini)
Ray-Ban Meta Gen 1 Yes Meta Llama models
Ray-Ban Meta Gen 2 Yes Meta Llama models
Meta Ray-Ban Display Yes Meta Llama models
RayNeo Air 3s Pro smart AR glasses No N/A
XRAI ar2 Captioning Glasses Yes OpenAI GPT models (requires user to set up OpenAI account)

Table 4: Comparing generative AI functionality and the models used across the glasses

To start, it is easy to focus on generative AI assistants, but smart glasses also have a host of other AI-powered features. To name just a few examples across the glasses we looked at – automatic speech recognition enables the use of voice commands in Meta glasses, text-to-speech allows printed content to be read aloud, and facial recognition helps users of Envision glasses identify people they know.

Saying that, most of our glasses do have generative AI capability too and there are a range of different AI models powering those features. Meta glasses unsurprisingly use Meta’s own AI models (the Llama model family). Envision’s Ally assistant, used in both of their glasses, actually draws on several models, including Llama, OpenAI’s GPTs and Gemini, selecting between them depending on the task given.

The value of conversational AI in smart glasses lies in multimodal interaction. XRAI for instance allows users to set up an AI assistant which extends the capability of the captioning glasses, allowing them to summarise conversations and ask questions. As the glasses do not have a camera, the AI can’t be used to analyse visuals, though. For those glasses with both cameras and generative AI assistants (e.g., Meta and Envision), users can capture images and ask questions that rely on visual context, such as requesting descriptions, directions, summaries or practical advice. Delivered through glasses, these multimodal interactions become a hands-free experience.

 

Displays, HUDs and augmented reality

Model Visual Display?
Envision Glasses No
Envision Ally Solos Glasses No
Ray-Ban Meta Gen 1 No
Ray-Ban Meta Gen 2 No
Meta Ray-Ban Display Yes
RayNeo Air 3s Pro smart AR glasses Yes
XRAI ar2 Captioning Glasses Yes

Table 5: Comparing which glasses have visual displays

There are quite different implementations of visual displays in smart glasses. Some include heads-up displays (HUDs) that present information at the edge of the wearer’s vision, while others aim for a more immersive augmented reality (AR) experience where digital content blends into the real world. HUDs alone are to some not considered a true example of AR, however we often see these terms used to describe products interchangeably.

There are many AR glasses available to purchase which are intended as a way to experience content in a more immersive way, or without requiring an external screen. This is how the RayNeo glasses operate, providing the user with a large virtual display which they can use for movies, gaming and more. These function primarily as a virtual screen for media consumption, rather than a tool for daily support.

XRAI’s captioning glasses make use of a HUD to provide the wearer with live captions in their eyeline, these appear in a single colour display (green). These allow the user to stay involved in conversations and other situations without having to look separately at captioning on a device. According to their website, XRAI look to be soon providing new models with full colour displays, more customisable captions and head tracking to allow for 3D captioning.

The Meta Ray-ban Display glasses, which combine a full colour HUD, AI assistant, cameras and microphones are therefore quite a unique offering currently. They were set to be available in the UK in early 2026, though recent reports indicate the international release will be delayed due to an increase in demand from US buyers. Notably, they also have a live captioning feature, including live translation (demonstrated in this clip). Meta’s future focus, alluded to in their “Project Orion” announcement, seems to be on developing glasses which achieve more of that true AR experience than the HUD experience.

 

Accessibility and inclusive design

Model Inputs Battery Life Prescription Lenses?
Envision Glasses Touch bar (tap/swipe), Voice control 6 hrs Yes
Envision: Ally Solos Glasses Tap input, Voice control 16 hrs Yes
Ray-Ban Meta Gen 1 Capture button, Touchpad, Voice control 4 hrs Yes
Ray-Ban Meta Gen 2 Capture button, Touchpad, Voice control 8 hrs Yes
Meta Ray-Ban Display Neuralink wristband, Touchpad, Voice control, 6 hrs Yes
RayNeo Air 3s Pro smart AR glasses Buttons and display on AR screen Not stated Yes
XRAI ar2 Captioning Glasses Buttons, Touchpad, Voice control 8 hrs Yes

Table 6: Comparing input methods, battery life and ability to use prescription lenses with the glasses

Many of the most interesting differences between smart glasses appear when accessibility is considered. Again, it can be important to note that products designed as assistive technology often incorporate features that are absent from general consumer devices. Envision’s glasses, for example, use audio cues to support navigation through menus and settings.

Other important considerations will include compatibility with other assistive devices. All of the glasses we reviewed support prescription lenses, although the details vary. Meta’s Gen 1 and 2 frames currently support prescription range (-6.00 to +4.00) and the Display glasses a more limited range (-4.00 to +4.00). Envision’s standalone glasses require additional frames as the base device (built on the Google Glass enterprise model) does not include lenses at all.

Hearing aid users may be able to connect their hearing aids via Bluetooth to certain smart glasses, though this is not likely to be possible through FM systems. Envision provide some advice on possible connections with some hearing aids but with the variety of hearing aids available and varying specifications they do not provide advice or guarantees for compatibility. Some providers recommend trial periods or have returns policies to allow users to test connections.

Accessing remote support via smart glasses may be another important function for disabled users. Envision glasses have functionality for the wearer to receive calls through the glasses and therefore can be used to phone a companion or a remote video interpreting service like AIRA to get support using the glasses camera. For Meta glasses users, a partnership with Be My Eyes allows blind and low-vision users to connect with their service, alternatively they can call personal contacts using Meta platforms like Messenger using the glasses.

Input methods also vary. Most glasses include some physical buttons or touch controls, alongside voice commands. Innovatively, Meta’s Ray-Ban Display glasses introduce a wrist-worn Meta Neural Band that interprets muscle signals to enable gesture control.

Battery life, too, is an important consideration when glasses are being used for assistive purposes. Most of the models we looked at included a charging case which charges the glasses while they are stored, but we have compared their base battery life on a single charge. Envision’s Ally Solos glasses were by far the longest at 16 hours, with most landing in the 6-8 hour range and Meta’s Gen 1 glasses trailing at 4 hours.

 

(Even) Cheaper smart glasses and knock-offs

Searching online marketplaces reveals hundreds of products sold as smart glasses, often priced between £20 and £100. Even with the reduction in prices of the models we’ve looked at, these even cheaper glasses are likely to be more accessible to students than high-end branded devices.

At the lowest price points, below £50, we noted that features typically include Bluetooth audio, microphones, AI assistants and audio recording. Slightly higher-priced models in the £50-£100 range advertised cameras and video recording capability.

Smart glasses are an area where powerful AI models could technically be integrated at a reasonable cost. These GetD AI Audio glasses for example, claim to offer integrations with ChatGPT-4o, Deepseek and Gemini 3 on glasses that cost just £55.99. The availability of powerful AI assistants, which can be integrated into cheap glasses could be adding a big boost to capability at a relatively low cost. This could mean that these cheaper options might be capable of providing useful functionality to the user.

However, there are many questions about quality, reliability and data handling with glasses being produced this cheaply. A broad look at customer reviews suggests significant variation in quality, and privacy policies are often unclear or absent. As with more expensive devices, users need to check terms carefully, but the risks may be harder to assess.

 

Conclusions

Looking across these products highlights how little the term “smart glasses” tells us on its own. Devices that share the same label may differ radically in capability, what data they collect, and what risks or benefits they present in educational contexts. For education, we see that the key challenge is not deciding whether smart glasses are allowed or not, but recognising the diversity within the category and responding accordingly.

In our second piece, we explore opportunities and challenges for smart glasses in teaching, learning and assessment.

Find out more by visiting our Artificial Intelligence page to explore publications and resources, learn more about our communities and sign up for our AI Literacy training.

For regular updates from the team sign up to our mailing list.

Get in touch with the team directly at AI@jisc.ac.uk

Leave a Reply

Your email address will not be published. Required fields are marked *