Cutting AI Down To Size
By Sarah Crespi, Sandeep Ravindran, Martin Enserink, Science.
Excerpt: TinyML (the ML stands for machine learning) is a low-cost, low-power implementation of AI that is being increasingly adopted in resource-poor regions, especially in the Global South. In contrast to the large language models (LLMs) that have dominated the news with their versatility and uncanny knack for humanlike expression, tinyML devices currently have modest, specialized capabilities. Yet they can be transformative. Murugan’s tinyML-equipped drones, for example, have been able to identify cashew leaves with the fungal disease Anthracnose with 95% to 99% accuracy. They should save farmers time they would otherwise spend looking for signs of disease themselves. And their ability to target treatments to diseased plants removes the need to indiscriminately spray pesticides on all the plants, which is both expensive and damaging to health and the environment. ...once the AI model is trained on a personal computer, it can often run for weeks on low-power tinyML devices powered by everyday batteries, sipping as little electricity as a typical laser pointer. ...TinyML - $2–$60 Cost per device (including sensors); Average power consumption per device ≤1–100 milliwatts. LLMs - $25K–$70K Average cost per AI chip, requires tens of thousands of chips; Average power consumption per AI chip: 700–1200 Watts...