Marketdash
Back to News

Meta's Smart Glasses AI Gets Crushed by ChatGPT in Head-to-Head Testing, Gene Munster Reports

MarketDash Editorial Team
11 days ago
Deepwater Asset Management's Gene Munster put Meta's AI-powered smart glasses through a rigorous test against ChatGPT, and the results weren't pretty. Meta AI understood 90% of prompts but delivered satisfactory answers only half the time, while ChatGPT nailed nearly everything at 98% accuracy.

Sometimes in tech, you need someone to just come out and say it: the emperor's new AI might not have clothes. That's essentially what Deepwater Asset Management's Gene Munster did this week when he published results from a systematic comparison between Meta Platforms Inc. (META) AI on the company's Display smart glasses and OpenAI's ChatGPT running on an iPhone.

The verdict? It wasn't even close.

The Numbers Tell a Brutal Story

Munster and his Deepwater partner Brian Baker put both systems through their paces with 50 identical prompts. The results paint a stark picture of where Meta's AI currently stands. Meta AI managed to understand roughly 90% of the requests thrown at it, which sounds decent until you realize it only delivered satisfying answers about half the time. ChatGPT, meanwhile, understood 100% of prompts and provided satisfactory responses 98% of the time.

"Meta AI felt like it was guessing while ChatGPT performed like an expert," the analysis noted. The testing revealed that Meta's assistant frequently struggled with basic comprehension and often ignored what was directly visible in front of the user.

Summoning Meta AI on the glasses proved "harder than it should be," according to the report. Instead of analyzing the live scene in front of users, the assistant had a frustrating habit of defaulting to generic, unhelpful replies. Simple tasks like describing an office environment or choosing between two visible items resulted in vague or completely unrelated answers.

The Cartoon Test and Other Failures

One particularly telling example came from what Munster dubbed "the cartoon test." When shown the same image, Meta AI simply labeled it without providing any meaningful insight. ChatGPT, by contrast, offered a fuller explanation of both the joke and its context, demonstrating the kind of nuanced understanding that makes AI actually useful in daily life.

The report did acknowledge "one bright spot" where Meta AI held its own: foreign language translation and short sign recognition. But that's a pretty narrow lane of competence for a product positioned as the future of wearable computing.

Why Munster Still Believes in the Vision

Here's where things get interesting. Despite these frankly dismal test results, Munster says he still supports Meta's nearly $18 billion annual Reality Labs investment. His reasoning? AI-powered wearables will eventually become a major computing platform, even if Meta hasn't figured out how to make them work properly yet.

Munster said he agrees with Meta CEO Mark Zuckerberg's vision for the space. "Now comes the hard part: making it work," he noted. Earlier this month, he expressed similar sentiments, acknowledging that despite impressive underlying technology, the Display smart glasses still fall short on the practical, day-to-day usefulness needed to appeal to mainstream consumers.

In other words, Meta might be building the right thing for the wrong moment, or more accurately, building something that needs a lot more work before it's ready for prime time.

The Hardware Lineup and Competition Ahead

Meta unveiled two new smart glasses models at its Connect 2025 event in September. The $499 Oakley Meta Vanguard targets athletes, while the $799 Ray-Ban Display comes with built-in AR capabilities. Munster previously described the lineup as offering the "best bang for the buck," though that assessment seems to focus more on hardware value than AI performance.

The competitive landscape is about to get more crowded. Apple Inc. (AAPL) is preparing its next wave of XR hardware, including future Vision Pro updates and a rumored smart glasses product expected to launch around 2027. Analyst Ming-Chi Kuo suggested back in June that Apple could ultimately surpass Meta's Ray-Ban smart glasses line, even with its characteristically slower entry into the market.

What Wall Street Thinks

Despite the AI shortcomings highlighted in Munster's testing, analysts remain generally optimistic about Meta's stock. The company holds a consensus price target of $826 based on ratings from 39 analysts. The three most recent ratings from Cantor Fitzgerald, Freedom Capital Markets and TD Cowen set an average price target of $776.67, implying a potential 22.55% upside from current levels.

The long-term bet here isn't about whether Meta's AI works well today. It's about whether the company can close the gap with competitors like OpenAI while building out the hardware infrastructure for whatever comes after smartphones. At $18 billion per year, Meta is certainly paying a premium to find out.

Back to News

Meta's Smart Glasses AI Gets Crushed by ChatGPT in Head-to-Head Testing, Gene Munster Reports

MarketDash Editorial Team
11 days ago
Deepwater Asset Management's Gene Munster put Meta's AI-powered smart glasses through a rigorous test against ChatGPT, and the results weren't pretty. Meta AI understood 90% of prompts but delivered satisfactory answers only half the time, while ChatGPT nailed nearly everything at 98% accuracy.

Sometimes in tech, you need someone to just come out and say it: the emperor's new AI might not have clothes. That's essentially what Deepwater Asset Management's Gene Munster did this week when he published results from a systematic comparison between Meta Platforms Inc. (META) AI on the company's Display smart glasses and OpenAI's ChatGPT running on an iPhone.

The verdict? It wasn't even close.

The Numbers Tell a Brutal Story

Munster and his Deepwater partner Brian Baker put both systems through their paces with 50 identical prompts. The results paint a stark picture of where Meta's AI currently stands. Meta AI managed to understand roughly 90% of the requests thrown at it, which sounds decent until you realize it only delivered satisfying answers about half the time. ChatGPT, meanwhile, understood 100% of prompts and provided satisfactory responses 98% of the time.

"Meta AI felt like it was guessing while ChatGPT performed like an expert," the analysis noted. The testing revealed that Meta's assistant frequently struggled with basic comprehension and often ignored what was directly visible in front of the user.

Summoning Meta AI on the glasses proved "harder than it should be," according to the report. Instead of analyzing the live scene in front of users, the assistant had a frustrating habit of defaulting to generic, unhelpful replies. Simple tasks like describing an office environment or choosing between two visible items resulted in vague or completely unrelated answers.

The Cartoon Test and Other Failures

One particularly telling example came from what Munster dubbed "the cartoon test." When shown the same image, Meta AI simply labeled it without providing any meaningful insight. ChatGPT, by contrast, offered a fuller explanation of both the joke and its context, demonstrating the kind of nuanced understanding that makes AI actually useful in daily life.

The report did acknowledge "one bright spot" where Meta AI held its own: foreign language translation and short sign recognition. But that's a pretty narrow lane of competence for a product positioned as the future of wearable computing.

Why Munster Still Believes in the Vision

Here's where things get interesting. Despite these frankly dismal test results, Munster says he still supports Meta's nearly $18 billion annual Reality Labs investment. His reasoning? AI-powered wearables will eventually become a major computing platform, even if Meta hasn't figured out how to make them work properly yet.

Munster said he agrees with Meta CEO Mark Zuckerberg's vision for the space. "Now comes the hard part: making it work," he noted. Earlier this month, he expressed similar sentiments, acknowledging that despite impressive underlying technology, the Display smart glasses still fall short on the practical, day-to-day usefulness needed to appeal to mainstream consumers.

In other words, Meta might be building the right thing for the wrong moment, or more accurately, building something that needs a lot more work before it's ready for prime time.

The Hardware Lineup and Competition Ahead

Meta unveiled two new smart glasses models at its Connect 2025 event in September. The $499 Oakley Meta Vanguard targets athletes, while the $799 Ray-Ban Display comes with built-in AR capabilities. Munster previously described the lineup as offering the "best bang for the buck," though that assessment seems to focus more on hardware value than AI performance.

The competitive landscape is about to get more crowded. Apple Inc. (AAPL) is preparing its next wave of XR hardware, including future Vision Pro updates and a rumored smart glasses product expected to launch around 2027. Analyst Ming-Chi Kuo suggested back in June that Apple could ultimately surpass Meta's Ray-Ban smart glasses line, even with its characteristically slower entry into the market.

What Wall Street Thinks

Despite the AI shortcomings highlighted in Munster's testing, analysts remain generally optimistic about Meta's stock. The company holds a consensus price target of $826 based on ratings from 39 analysts. The three most recent ratings from Cantor Fitzgerald, Freedom Capital Markets and TD Cowen set an average price target of $776.67, implying a potential 22.55% upside from current levels.

The long-term bet here isn't about whether Meta's AI works well today. It's about whether the company can close the gap with competitors like OpenAI while building out the hardware infrastructure for whatever comes after smartphones. At $18 billion per year, Meta is certainly paying a premium to find out.