Tiny AI? What’s that? Small Robot? Something futuristic? Believe it or not, Tiny AI is already here – on your phones!! Siri and Hello Google are prime examples of Tiny AI. When Siri was first introduced, it needed internet to work – but now – it can work without internet as well – a good example of how AI can work on “edge” devices.
Tiny AI is currently a big focus for AI researchers. Tiny AI refers to the new model of Artificial Intelligence or ML that makes use of compressed algorithms to minimize the usage of large quantities of data and computational power. Tiny AI compresses Artificial Intelligence algorithms such that they fit and run within microprocessors on consumer or the Internet of Things (IoT)-enabled devices. This emerging trend in Artificial Intelligence also aims to improve the sustainability of artificial intelligence.
Where can we see Tiny AI being used?
- Quality checks
Tiny AI enabled machine vision can allow machines to “see” the products on the production line and spot any imperfections. The system can then recognize defects, mark them, and send alerts.
While conventional industrial robots require being specifically programmed to carry out the tasks they were created for, but Tiny AI-powered robots can interpret CAD models, which eliminates the need to re-program their movements and processes every time a production line needs to produce a new product. In 2017, Siemens developed a two-armed robot that can manufacture products without being re-programmed.
The Autonomous car that the world is anxiously waiting for to appear on our roads, is a great example of Tiny AI coming into our day to day lives. Tiny AI will not only take care of navigating through traffic but will also take care of things like monitoring drivers’ alertness and health, controlling car entertainment systems etc.
As one can imagine, there are many use cases where Tiny AI can and will play a huge role in our near future. Tiny AI, in addition to the obvious benefit of being “portable”, there is the significant environment impact to consider.
Environment impact of AI?
Many people don’t associate AI with impact on environment – but the fact is, training sophisticated “full blown model” AI like Google’s Bert and Microsoft’s Turing Natural Language Generation model consumes a huge amount of energy. It is estimated that the carbon footprint of single training run of some of these AI models generates almost 5 times as much as the lifetime emissions of an average car.
Tiny AI is here to stay – Hey Siri – what do you think?