The AI Lab PrismML, co-founded by Caltech researchers and advised by UC Berkeley’s Ion Stoica, has unveiled a version of its minuscule language models for smart glasses, powered by Qualcomm’s Snapdragon chips. At the Snapdragon Summit, the chipmaker demonstrated PrismML’s 1-bit Bonsai LLM, optimised for local processing on AI smart glasses running the Snapdragon AR1 Gen 1 Platform. This model, a 2-billion-parameter beast, retains nearly all the performance of larger models, allowing wearers to query objects in real-time.

Their aim is clear: to democratise AI by running it on devices, optimising computing resources. This move represents a step towards a world where AI is decentralised, challenging the dominance of proprietary labs. Yet, no smart glasses running PrismML have been announced, leaving questions about adoption and integration.

PrismML’s technology could transform how we interact with technology, blending convenience with privacy, but it will take more than a new model to convince the world.

Their ambition is to create open-weight AI that makes better use of existing computing power, reducing reliance on cloud-based services. This could be a game-changer in the tech world, offering a viable alternative to the privacy concerns associated with large tech companies.