Meta's Smart Glasses Might Make You Smarter (www.wired.com)

At Meta Connect, Mark Zuckerberg doubled down on a vision that AI-enabled smart glasses will confer a “cognitive advantage,” but Meta’s live product demos highlighted how far the tech still has to go. When a chef used the “Hey Meta” wake word, hundreds of distributed demo units all activated and began responding — an issue Meta CTO Andrew Bosworth later attributed to the company’s own AI instances effectively DDOS’ing one another. Other demos suffered video-call failures, lags and repeated commands. Meta’s new Gen‑2 Ray‑Ban models are more fashionable and have sold over 2 million pairs, but they’re bulkier, produce distracting on‑screen notifications, and expose the limits of current ASR/NLU reliability and low‑latency distributed systems in crowded real‑world settings. For the AI/ML community the keynote is a useful case study: scaling multimodal assistants to many co‑located devices raises system‑level problems (wake‑word collision, resource contention, latency) as well as ML challenges (robust speech recognition in noisy scenes, context‑aware dialogue, on‑device inference to protect privacy). It also underscores human‑factors constraints — attention, social acceptability and accessibility tradeoffs — that matter as much as raw model capability. Practical fixes (conversation detection to mute notifications, better gesture/UX design, edge optimization) will be critical if smart glasses are to move from awkward demos to genuinely augmentative, trustworthy tools.