A Thesis on AI Hardware
What are the future prospects for AI hardware? Unless its sophistication increases, very limited probably.
The truth is this: advancements in AI software technology are continuing daily and at a breakneck pace – and there has yet to be a must-have piece of hardware that effectively makes use of these on the consumer side. The cellphone (later the smartphone) became integral in the lives of people by providing a type of spontaneous and instant interaction that had previously just been envisioned. And there’s no disputing that robotics have helped to transform manufacturing, in ways both positive and harmful. For instance, the implementation of robotics has robbed workers of jobs. Many of the jobs taken were good jobs, jobs that provided what would be considered a middle-class wage. On the other hand, robots have taken over tasks that were physically challenging (dangerous even) or in some cases impossible for flesh and blood workers to perform.
Why is the question of AI hardware’s future so relevant now? While it passed the notice of the general public, a once-ballyhooed robot maker (that boasted the attention of Apple no less – I mean could there be better sponsorship in the world of electronic gadgetry – or in the tech world in general) has elected to draw in the shutters on its once-promising operation. The name of this defunct startup is Anki. It’s one that folks in the world of AI certainly know. Likely because of its association with Apple, this startup has been on the radar really since its 2010 inception (in its existence Anki was profiled in a number of publications, among them The Wall Street Journal, publicity which comfortably eludes most startups). In the end, it wasn’t enough.
Anki specialized in what could be considered companion robots (and the startup raised an astonishing $200 million plus to do so). Unfamiliar with companion robots? That wouldn’t be a surprise; they haven’t exactly taken the world by storm. I’ve heard them described as “electronic pets” – a fair portrayal since they’re designed to interact with us (theoretically even interpreting our feedback) in a way reminiscent of our furry friends. At the time of its shuttering, Anki’s top product (courtesy of the company website), was a machine called Vector, in appearance a miniature drone (probably just small enough to fit in an average person’s palm). Vector maneuvers on treads like a bulldozer’s (isn’t an airborne drone) – could be the little brother of one of the drones deployed by a bomb squad Hollywood style. And what does Anki’s technological marvel, Vector, do exactly? It performs useful tasks like keeping its companion (i.e. you) apprised of the weather (I’m completely serious), or capturing that “perfect” picture at a moment’s notice, or preventing your dinner from overcooking with its built-in timer function.
No fooling. This is what Anki’s Vector would have done for you – at the bargain price of $250.00 for the unit (and more for the unit’s accessories).
This is the difficulty confronting not just Anki but its multitude of rivals who are working on smart robots to interact with human beings (to this point, that interaction appears to harbor no end game, it seems to offer no particular human benefit). Right now the usefulness, the utility of these creations simply isn’t there. For a piece of AI hardware to succeed in the consumer marketplace, people must conclude that it is indispensable, that it’s something impossible to live without (following the paths of historic inventions like the automobile and the original Ma-Bell telephone, later the cellphone (and from there the smartphone), the radio, the television, our personal computers and now our laptops – and a host of other massively influential products over the years. A companion robot that offers nothing but companionship may prove to be an interesting short-term conversation piece (if you’re lucky and you entertain a lot) – before it’s hidden in a box and pushed onto a closet shelf (having proved a frequent headache, a genuine nuisance, or both). If you watch old sci-fi productions from the nineteen fifties and sixties (I have a thing for old movies), you’ll see featured robots vacuuming hallways, cleaning dishes, responding to doorbells, walking dogs and operating motorized vehicles (and by the way, a self-driving car would likely be considered a robot even if it doesn’t particularly look like one); robots capable of performing these everyday tasks effectively are what we need to see as consumers if we’re to step in and help save the ailing (and for that matter the ailments are worsening) AI hardware industry.
A robot the likes of the typical sci-fi creation could have tremendous value to consumers – if it was affordable. A viable robot, one capable of performing household chores, could command (you would think) at least several thousands of dollars – consider for a moment all of the other things that we consumers spend copious amounts of money on to make our lives easier, to improve our leisure or to create peace of mind – things like lawn tractors, home security and backup systems, home theater and entertainment setups, wooden decks outside our sliding glass doors – and plenty more. I have nothing against the ownership of these, but imagine for a second just how much more useful an entity of some kind – we don’t know for sure what a working household robot would look like (would it resemble a human, would it look like R2D2, would it ape a treaded drone dispatched by a Hollywood-depicted bomb squad to detonate a terrorist device – this would be an entity capable of dusting and polishing, of making a bed in the morning, of cleaning a toilet or sink, of safely putting away clean dishes, of otherwise erasing all the little messes we leave behind in our daily travels, the messes that our kids and our pets generate with such effortlessness; it would be capable of taking over without complaint the dingy deeds of modern living that we find laborious, inconvenient, or downright unpleasant (I’ve yet to meet anybody who was chomping at the bit to get started cleaning a toilet). It seems that our advancements in AI hardware technology, while impressive in their own right (I don’t want to diminish the great work that’s been done by companies like Anki), haven’t taken us close enough (so far) to this goal’s resolution to help the ailing AI hardware industry.