AI News Monitor

In a world dominated by power-hungry artificial intelligence (AI) models like ChatGPT, a quieter, more sustainable revolution is taking root in the Global South. TinyML, a low-cost, low-power AI technology, is enabling researchers and innovators to tackle local challenges—from agriculture to healthcare—without the need for expensive infrastructure or internet connectivity.

TinyML devices, often no larger than a pack of cards, run on microcontrollers that consume minimal power, allowing them to operate for weeks on simple batteries. These devices are being used to detect plant diseases, monitor heart rhythms, track wildlife, and even classify mosquito species by the sound of their wings.

One such innovator is Bala Murugan, a computer scientist at the Vellore Institute of Technology in India. Drawing from his family’s cashew farming background, Murugan developed a drone equipped with a tinyML model to identify fungal diseases in cashew leaves with up to 99% accuracy. This system minimizes pesticide use, saving farmers time and reducing environmental harm.

Similarly, João Yamashita, an electronics engineer in Brazil, created a tinyML device to diagnose diseases in coffee plants. His $20 device, which doesn’t require internet access, has achieved 96% to 98% accuracy in identifying diseases like rust and leaf miner moth infestations.

TinyML’s applications extend beyond agriculture. In Malaysia, researchers are using the technology to monitor plastic pollution in rivers, while in Brazil, it’s being used to detect abnormal heart rhythms. In Kenya, tinyML devices are helping classify malaria-carrying mosquitoes, speeding up disease control efforts.

The technology’s appeal lies in its affordability and accessibility. TinyML devices cost between 2and2and60, compared to the tens of thousands of dollars required for high-end AI chips. They also consume as little as 1 milliwatt of power, making them ideal for regions with limited resources.

Despite its modest capabilities, tinyML is gaining traction globally. Researchers predict that the number of tinyML devices shipped annually could grow from 15 million in 2020 to 2.5 billion by 2030. Workshops and courses are spreading across the Global South, empowering local communities to develop solutions tailored to their needs.

“TinyML will enable AI to go everywhere,” says Yamashita. As the technology continues to evolve, it promises to bring transformative change to the world’s most resource-poor regions, proving that sometimes, smaller is better.