Edge computing is the way forward, but are we ready to surrender control to corporations.
Intel’s research on self-driving cars or autonomous vehicles highlighted that a car would generate 1GB of data every second and would require computation of this data in fraction of a second for decision making. But are cloud networks fast enough to receive 1GB of data within a second, compute possibilities and relay it back. Otherwise, between the time you ask your car to switch on the light and the time it crashes into a pole, there would be little left. Systems, no doubt, are getting faster, but a new technology is becoming pervasive. Edge computing is fast becoming a base for most IoT devices. So, as users send more data to the cloud, companies are trying to put more computing power in devices.
Artificial intelligence on edge is gaining momentum. While the concept is not new, given the focus on privacy and security, it may become more critical as privacy scandals get uncovered. The most straightforward calculation of computation on edge is your mobile phone. With certain compute capacity, your phone stores specific data which it does not even share with the central server—fingerprints and face scan. The earliest fears about these technologies were that companies would get a hold of personal data and may be able to build elaborate profiles. The next time you enter a store and the cashier knows your name, you would dread Samsung or Apple.
So, what phone manufacturers did was assure you that the data shall always stay on the phone and never make it to Apple’s database. The phone will decide on encoding and decoding a fingerprint. Initially, phones made mistakes, but as technology gained pace, they got better at it. Siri or Google, on the other hand, rely on the cloud. Once you ask a question, the request is sent to the cloud where the AI determines what it means and sends back the answer, which is relayed to you. A couple of years ago, Amazon announced that it would be creating a custom chip for Echo so that not every piece of information is gathered on the cloud, but is calculated and determined within the device. So, while Alexa would still rely on the cloud for specific commands, it shall also be customised to each consumer to offer a more personalised experience.
All this seems to complicated. So, why are companies going through such an elaborate exercise? Companies also realise that not everything can be transferred to the cloud. So, while it is easy for Google to store basic things, it would be near impossible to calculate all variables in a self-driving car from a cloud and generate data. Meanwhile, also compute how the vehicle interacts with other cars and traffic systems. Edge, thus, becomes the only solution. This way, a car would not have to send everything to the cloud. Certain decisions will be taken by the computer onboard. A security cam, for instance, can decide which information to hold and which to dismiss.
The information it holds can be transferred to the cloud. A refrigerator can determine the contents and decide on how much temperature it wishes to set, for intimating the supermarket that you have run out of milk, it still needs the cloud.
Computing on edge can take various forms. But one thing it still does not address is the control over data. While companies may talk about privacy, they still want data from you. And, as more devices get into the IoT range, it would benefit from sharing this data. Edge means personalisation, but would companies be willing to give up control. While Amazon loses control over certain data, it shall want access to more devices.
Moreover, the consumer would no longer wield control over devices. For instance, computers gave you an illusion of choice about downloading updates. Edge devices won’t. They would upgrade themselves with little user control. The only way companies can drive adoption is by giving up on control, but as more devices get connected, they would have more leverage. Edge, thus, needs more solutions. Blockchain is one answer.