https://vmblog.com/archive/2021/12/23/brainchip-2022-predictions-...

https://vmblog.com/archive/2021/12/23/brainchip-2022-predictions-2022-a-breakthrough-year-for-ai.aspx


2022 a Breakthrough Year for AI

Each year I am privileged to be asked to comment on the future of AI technologies, their progress and their promise in the year to come. In 2022, I believe we will see more progress, and more promises realized, than any year so far. Here is what I and my colleagues in AI expect to see:

1. AI processing will increasingly move from the cloud into edge devices with the availability of the Akida neuromorphic processor. Distributed processing has many advantages, including the reduction of total power consumption. Cloud datacenters emit as much as an estimated 600 megatons of greenhouse gasses a year. Distributed processing could reduce this figure considerably, because neuromorphic processing of CNNs on an SNN platform uses 97% less energy. AI devices will work independent of the cloud and the internet in 2022. This is similar to what happened in the 1980s: before 1980, data was processed at central mainframe computers. The IBM PC changed that to distributed processing.
   
2. The rigid "train once and stop learning" rule of CNNs will become a limiting factor in CNN development. Real-time learning of new objects and faces will become more important as AI devices will need to have the capability to be personalized by the end user without extensive retraining. The neuromorphic processing of CNNs on an SNN platform offers this capability.
   
3. The AI field will continue to expand beyond visual recognition to include all the human senses, including odor and gas recognition, taste classification, voice identification, vibration analysis for early fault detection, and other ‘expert systems.'
   
4. AI will become more commonplace in everyday products; refrigerators that can smell if any food is about to go off to prevent food poisoning, washing machines that monitor waste water for pollutants, TVs that understand our voice commands, cars that recognize their drivers and make sure the driver is awake and paying attention, hand-held medical diagnostic monitors that detect diseases from a single breath, smarter ADAS that continues to learn and become a better driver with experience, and more use of AI in industrial sensing to create a safer work environment.
   
5. AI tools will become easier to use, not requiring any specialist knowledge, whereby the development systems will hide the complicated ‘expert' work under a user-friendly graphical user interface, much like what Apple did for computer operating systems when MS-DOS was the common norm.