Alexa, Google Assistant, Cortana, Google Now and Siri have more in common than being A.I.-based systems that are made to assist people digitally. They’re all platforms; they can be put on just about any device and follow a user from device to device and place to place, learning more about them and how to cater to their needs in a contextual manner and ensuring that they’re always available to the user. While some of these assistants are better at that than others for the time being, the reaction to Google’s I/O announcement about Google Assistant has made it quite clear where the focus lies for the future. With Amazon Alexa picking up steam, mostly in the form of Amazon’s home-based Echo, and Google Home coming into the market soon, some may say that the technology has found its ultimate physical form and best use case. That, however, is probably not true.
Before going into how digital assistants will find themselves at the helm of a technological revolution, it may be a good idea to recap what makes them so special and what is going to make them much more special in the very near future. That secret sauce is machine learning. Thanks to breakthroughs in, essentially, the way computing works, things like neural networks and more powerful processors have enabled computers to learn in much the same way humans do. While they normally have to be spoon-fed information to learn it and use it in context, certain setups allow an A.I. to learn naturally. This requires a lot of backhaul, but this is power that Google and Amazon, among others in the field, are more than able to provide. With machine learning and a contextual A.I. in place, digital assistants can do things like figure out that you’re near your favorite restaurant and that it’s been around 6 hours since you last went into a store or restaurant, then show you the restaurant’s menu. With current integration technologies, the assistant could also check your bank balance against recurring bills and statements to see if you can afford to eat at that restaurant and, if not, possibly suggest cheaper alternatives in the area.
Another key component in the future of A.I. and thus the future of these assistants is neural networking. While a full discussion of how neural networking works is best saved for another time, the long and short of it is that computers can combine their power to run an A.I. and extrapolate functions a single machine couldn’t perform on its own. Each node in the network acts like a neuron in the human brain, giving an A.I. a framework backend that’s very similar to a human brain and allowing them, in essence, to think. This ties into the discussion with the assistants not only by virtue of their probable future use of the technology, but in the way that use could be facilitated. On top of the cloud servers of the company who makes the assistant a user chooses, the assistant could work across multiple devices that a user owns and use them for small-scale neural networking, albeit for low-level functions at a low speed. This is another good reason that a universal approach for virtual assistant A.I. is the most likely solution for the future.
Home hubs, in their current form, represent the ‘nucleus’, if you will, of the future framework of such technologies. Currently, only Google and Amazon have home hubs either in the works or on the market, but others could jump in the game in the future. A home hub can obey commands, link to devices all around the house and, of course, act as an IoT control hub. If a future where you no longer interact physically with a device, but use an A.I. to connect with it ever comes to pass, it’s a safe bet that home hubs will play a huge role as a base of operations of sorts for these A.I.s. Cars, wearables, new forms of smartphones and a million other types of devices that exist now or may be dreamed up in the future could all become nodes for an A.I. centered around a single user; omnipresent, omniscient and always willing to help in the most contextually suitable way possible. While that may sound scary, there’s a distinct possibility that, given that security and the hardware can keep up, this is exactly where assistant A.I. could be headed.