There will come a time when even the simplest of devices will become smart. From the sprinklers in your garden to the medical implants in your body, everything will respond to the environment in the time to come, courtesy of artificial intelligence. So, if you are worried about the Terminator knocking on your door say, in 2045, you may not be wrong.

#Microsoft's precise goal is to lace AI into the most unimaginable and innate of devices. By successfully running AI on the Raspberry Pi, which is a portable microcomputer with relatively low-key functions, it has made a commendable step in this endeavor.

Artificial Intelligence reduces dependency on cloud-based algorithms

Currently, the smart technology involves relaying data over sensors using the Internet or other wireless means and then relaying back the decisions.

Advertisements
Advertisements

This will change with the advancement of AI. The idea behind this was generated at two of Microsoft's labs, one in Redmond and the other in Bangalore, India.

What happened is that Ofer Dekel, an employee in the AI department at Redmond, got irritated by rodents like squirrels, who destroyed gardens by feeding on flower bulbs and seeds. He, then, programmed a computer sensor system to sense out these rodents. Once detected, the system would then instruct the sprinklers to go live, spreading water and chasing away the rodents. The programming was done on a Raspberry #Pi3 microcomputer.

The cost of such AI interfaces is still high for mass consumption

The current cost of technology pegs uses of AI on simple devices like sensors and sprinklers against us. Installing advanced chips or even cloud tech on such devices is costly and can only be pursued currently for recreational purposes and hobbies.

Advertisements

However, this example of a Raspberry Pi3 being able to handle such code is a positive move towards the inevitable. Systems with only few KB RAM need compressed parameters much lower than the general 32 bits. This was done for the Raspberry Pi. Another technique is the filtering down of algorithms to the maximum detail to remove redundancies. This enabled the team at Microsoft to make image detection 20 times faster with no loss in accuracy.

For stable and mass market use, new AI techniques need to be invented for devices that are low in memory and power. With the ongoing exponential rate of technical disruption, we are bound to see newer technologies in the future, and it seems it will not be long until each equipment in our environment will start responding to one another and us.