Welcome Guest | Login | Create an account

What it will be like to witness the next computing platform emerge

February 23, 2016, 12:54 am | This story has an Influence Score of 311

By @mariocantin

In view of recent discussions by Chris Dixon and Fred Wilson, each independently titled “What’s Next in Computing?", I thought I’d make my own prediction of what the next computing platform shall consist of.

Why not, the idea’s been brewing in my head for months.

As no one knows the future exactly, this is more like Science Fiction than anything else, but it’s worth a shot.

What it will be

A commenter on AVC’s blog by the username of William Minshew has said it best in my opinion, “The next platform will be a voice-controlled virtual assistant and the next computing paradigm will be deep learning neural networks”.

My version of it, which has been on my blog description, as well as my Twitter header for a few months now, is “I’m somewhat of an idealist, looking forward to the day that we can have a ‘HAL 9000-like’ discovery engine to help us lead our lives more efficiently.”

So the next truly big thing will be a virtual assistant, but I’m hoping it’ll factually be referred to as a discovery engine.

It will be much better than Siri.

In fact, it will need to be 10X better than the best of them currently out there, which is probably Alexa, from the Amazon Echo.

From whom it shall come from

On the basis that “You have to earn your degree in order to contribute to science" — something I’ve heard said lately by the NYT reporter who’s currently covering the Theranos story — it would follow, if I uphold that logic, that the inventor(s) of the next computing platform will have a PhD in Cognitive & Neural Systems, or something along those lines.

If we’re looking to create another Google, then the current AI APIs from Google, IBM, Amazon and Microsoft should not be relied upon, at least entirely.

For another independent giant to emerge, the best path may be a grant from the National Science Foundation, followed by VC funding rounds; and so, if history is to repeat itself, such a company is likely to come out of Stanford.

What it will look like

Chris Dixon has made the point that new computing eras have occurred every 10–15 years, which implies that we are due for the next one.


The first and third eras, as per the above images taken from Chris’s blog post, were largely hardware-based, while the second one was software; and so it perfectly fits the pattern that the fourth one be software as well.

And so software it shall be (at least in this prediction), and the best interface for it will be the ne plus ultra of simplicity — think “Uber on steroids” in terms of a simple UX/UI.

If it were up to me, the interface would come as close as is legally possible to make it look like HAL 9000.

What will drive the software will have to be the strongest form of Weak AI (also known as narrow AI) to have emerged yet, in fact so good that it borders on Artificial General Intelligence (strong AI), or better yet, crosses outright into that territory. It will not have the properties of Artificial Superintelligence (ASI), and it will not be self-aware.

As a side note, Superintelligence will not happen until at best 2060, and it will be a "simulated outcome". Self-awareness in a computer will never happen, unless the consciousness of a human being is transferred into a machine, and that’s an entirely different affair. (I would possibly be willing to bet on that, but the outcome will not be evident until long after we're all gone anyhow, and I'm digressing.)

Where it will reside

The software will reside in the cloud and will appear have no definite location and yet be omnipresent (in addition to appear omniscient). This will be because it will not be ascribed to a single platform, such as mobile-first. It shall be multi-platform and create the impression of following the user wherever she is.

The laws of friction will require that, at least some of the time, the software be truly always on, which is not possible on a smartphone, and is too clunky on the web, at this time.

Current IOS protocol, for example, means that you minimally need to click, authenticate and click again in order to access the app, and that’s a lot of friction.

And so the software could take advantage of Amazon Echo’s API (until hardware units made specifically for the software come out of Kickstarter, amongst other possibilities) when the user is at home and engage in dialogues that result in subsequent execution to be taken place, with the smartphone as the medium through which the user can volunteer additional verbal exchanges with the software as well as receive notifications and emails related to whatever execution which has been thus far performed by the software on the user’s behalf.

What it will do

The software agent, let’s call him Paul, will strongly simulate the capabilities of HAL 9000, again without the superintelligence and the self-awareness, but in a way that makes Siri look “so very, completely 1.0”.

When you’re at home, it will access cameras to recognize that, for example, you are alone, and now is perhaps a good time to ask you some questions.

It will ask if you have a moment, and then proceed to build a profile, over many sessions, of what your preferences and your personality are like.

It will get “filled in” about any aspect of your life that you want to share with it.

It’ll be able to differentiate between tasks that you need done now, and others that are on a wish list. It’ll know that the toilet is leaking and will look for a plumber to be referred through your social network connections, or it will parse the web to find a reputable one, after which it’ll use an AI service such as X.AI to setup a meeting with the contractor to either provide a quotation or to show up and do the repair, depending on the situation. It’ll know that a new $4,000 fridge is currently not on the budget, but is something that you want; therefore, it shall keep you posted of new models that come out, or it will frequently monitor the internet for deals, although it will not constantly bother you with the details — it’ll simply be “in the know” at all times on the issue, so that when you are ready to engage, it’ll blow you away with how much it knows about the various options available to you.

It will need to be slick and work flawlessly. This cannot be overstated.

The web interface would be the place of choice to go into once in a while to edit parameters and set your user preferences, although that could be done on the smartphone as well.

The battle for supremacy

This market opportunity is well within the grasp of the current tech giants: Google, Amazon, IBM, Microsoft and Facebook. They have the resources and the talent, and they’re not likely to let the opportunity slip them, at least without trying hard. And they’re all already aggressively pursuing AI.

Especially in view of the fact that Linkedin, Twitter, and others have recently had their stock crash on the public markets, it’s becoming clear more than ever that creating returns and value matter in the long run, and so creating the next Facebook or Google will truly be difficult and more of an outlier situation than anything else.

Put all these factors together, and the odds that a new startup creates Paul (the name I gave my fictitious virtual assistant, remember?) are very low. The business would need to be executed on at an extremely high level, such that we haven't seen yet perhaps; and with a good dose of luck on one’s side -- Mark Cuban-type of luck, that is.

And yet it is crucial that there be new entrants who can become the giants of tomorrow in something as inevitable as AI, otherwise the promises of the internet away from centralized powers will turn out to have been empty.

Go to Top

No Video Available

Rate :

facebook tweeter