Why Cognition-as-a-Service is the next operating system battlefield

42 Comments

The Semantic Web may have failed, but higher intelligence is coming to applications anyway, in another form: Cognition-as-a-Service (CaaS). And this may just be the next evolution of the operating system.

CaaS will enable every app to become as smart as Siri in its own niche. CaaS powered apps will be able to think and interact with consumers like intelligent virtual assistants — they will be “cognitive apps.” You will be able to converse with cognitive apps, ask them questions, give them commands — and they will be able to help you complete tasks and manage your work more efficiently.

For example your calendar will become a cognitive app — it will be able to intelligently interact with you to help you manage your time and scheduling like a personal assistant would — but the actual artificial intelligence that powers it will come from a third-party cloud based cognitive platform.

Cognitive apps will not be as intelligent as humans anytime soon, and they probably will not be anything like the 20th century ideas of humanoid robots. But they’re going to be a lot smarter than the software of today.

Cognition in the clouds

Clouds SF

But the key is that the intelligence that powers cognitive apps will come from cloud based platforms that host their brains — the apps themselves won’t really have to be that smart on their own. Which means that truly vast, always increasing, intelligence will be available via APIs to all kinds of apps, and right into the full range of consumer appliances, devices and even the Internet of things. All apps and even things will start to become cognitive.

Even in the last few months several interesting announcements were made that all signify this trend:

  • The startup, Vicarious, has developed a new form of AI that is capable of reading CAPTCHA images, the most widely used test for differentiating human and computer actions online.

  • Next IT announced the Alme platform for virtual healthcare assistants, furthering the development of intelligent virtual assistants with domain-specific expertise

  • Google is finally starting to make search smarter by incorporating more contextual conversational capabilities for queries, and even challenging Siri directly within iOS.

  • Stephen Wolfram announced the Wolfram Language, which models the world and combines both programs and data — what he calls a new “language for the global brain,” — that will essentially be able to weave sophisticated computational knowledge into everything.

  • And finally, IBM announced it is going allow third-party developers to build cognitive apps that leverage cognition hosted in the cloud on Watson.

All of these announcements foretell the development of platforms that will allow apps and services to all function more intelligently and intuitively. In fact, the coming competition between different CaaS platforms may be the 2015 – 2030 equivalent of the operating systems wars of the 1980s and 1990s. This means CaaS platforms are the strategic high ground in the battle to own the future operating system of the Web and mobile applications — the operating system of the global, networked brain itself.

The new OS battle

Toy soldiers line up for battle

We already see growing signs of heated competition between Apple and Google to make smarter mobile virtual intelligent assistants. How long before both companies open up API’s to their CaaS platforms to the rest of their ecosystems?

App developers will soon need to choose which CaaS ecosystem to build on — Google, Apple, Facebook, Microsoft, or maybe even Wolfram’s new Wolfram Language ecosystem.

In the long-run however, a more vendor neutral cognition platform may emerge as the winner: one that is more like Amazon Web Services in that it just provides the underlying service and doesn’t compete with third-party apps that use it. This could come from Amazon, or Wolfram perhaps. CaaS platforms may eventually even be open-sourced and made widely available — perhaps via a Linux equivalent for the cognitive operating system era that might borrow from many of the original ideas and standards of the Semantic Web.

Bringing computer science smarts to the masses

Home appliances, connected home, internet of things

The original goal of the Semantic Web was to provide an open framework for apps to integrate and reason across any data in any repository, anywhere. But the open standards that were introduced by the WC3, such as RDF and OWL have proved to be unwieldy and difficult for non-computer scientists to grasp.

Where CaaS comes in is that much of the reasoning capabilities that Semantic Web hoped to enable could now become simple APIs in the cloud that any developer can use, without needing a PhD. As we head toward 2020, CaaS will make the world more helpful. The cognitive operating system will reach out and connect our bodies and even REACH into them via augmented reality devices like Google Glass, and the quantified self movement.

Everything is going to get smarter. Your phone, your calendar, your watch, your radio, your TV, your car, your refrigerator, your house, your glasses, your briefcase and clothing. The vast cognitive capabilities of the global CaaS providers will be cheap and available via APIs to every device from the nano scale up to the giant global applications and services.

A world that is constantly thinking and trying to help out with every little task, could be more efficient, but it could also potentially feel pretty annoying, obsessive compulsive, and invasive to us. But a world where everything is smart probably will be quite normal to our grandkids. In their world cognition as a service could be as ubiquitous as electricity is for us today.

Nova Spivack is the co-founder and CEO of Bottlenose, an angel investor, and a noted technology futurist.

42 Comments

Stefan

I’m glad to see good old-fashioned AI come back in new clothes. Privacy will be the HUGE issue — we could easily slip into a command and control world not worth living in here. Every engineer working on semantic technologies hopefully will remember this.

Michael Holmes

Thanks for the thoughtful article. I’m Program Director at IBM Watson Solutions and am very excited to see this discussion on the value of establishment of ecosystems of organisms with shared interest in embedding cognitive capabilities into a new generation of apps. While the term “cognition as a service” may be partially applicable, our team is mostly interested in the enthusiasm that is building around a new generation of cognitive apps…. regardless of what term is used to describe the underlying engine. Thanks again!

Simon

In light of recent developments: What about security and privacy? Will CaaS become yet another venue by which governments (and advertisers) will use to monitor people?

ingmar

Great article, nice insights and enough food for thought indeed, yet i think all this change is not happening so we can be more efficient, i think we are striving to be more effective!

ogenex

Not a bad article, not because it provides any new or deep insight but purely because it draws our attention to the natural direction in which our computational brains are heading.

Reading some of the comments prompts me to advocate for a less structured paradigm, one that does not require concepts and relations to be labeled- but a system that is able to ingest free and open unstructured text and data sources. No doubt the likes of Wolfram and Google labs are well on their way to realising such systems. What is clear is that next generation programming languages must evolve to natively handle knowledge- both in terms of recognising and parsing sources of information, representing the content, and inferring new knowledge in context. The ability to process natural language is a large part of the equation.

Lance Zant

Exciting stuff, especially if we’re prepared to add a little patience and perspective. The patience comes in realizing that we chronically over-estimate what we can accomplish soon, but just as chronically underestimate long term changes. That makes who’s in the market now much less important than recognizing the real direction things are going. If the current bunch turn out to be the VisiCalc of CaaS, so what?

As for perspective, we need to consider possible down-sides, especially that of misuse by both crooks and the powerful. It might help to recall that at the dawn of the “Atoms for Peace” nuclear power era, they said electricity would become so cheap we wouldn’t need to meter it.

jonathanross

Great stuff, fascinating. Nothing to add about Microsoft’s supposed Cortana Project? I know information on it is vague but it’s also worth keeping an eye on.

Jason

Never mind smarter things, we should be working towards the politics and greed that is making energy so expensive. At this rate the civilized world as it is today won’t be able to afford to power the smart devices and they will be useless then.

Ravi

Kingsley is spot on – this is just creating buzz around technology so someone somewhere can sell something or help his friends sell something. Even the word “Enterprise” and the whole host of technologies that get labeled with it have nothing to do with the academic field of Computer Science – which is where most concepts are born borrowing from other areas of academia, especially mathematics.

Nevertheless, its is always great to see the real world use this stuff to solve problems in really novel ways. Just don’t pay too much attention to the labels and politics, especially if you’re in the field.

RudeBaldGuy

I like the concept of “the brain in the cloud”. Maybe we could call it SkyNet. What could go wrong? I’m sure NSA and DARPA could put it to use for the good of all human/robot-kind.

On a more mundane note, this guy says they have finally figured out how to read captchas. Great, now a moderately successful security tool is binned. Would it be surly of me to ask whether they could put their gigantic brains to work fixing the dreadful security issues of the current Internet instead of gleefully shredding the pathetic sticking plasters we have to use to cover these shortcomings? Just asking.

Some things to work on:
DDOS
Workable hardware identity tokens
SQL injection
XSS
Man in the middle

Clearly not as sexy as putting Siri in my dishwasher to nag the shit out of me, though. I get it.

Brian Buckley

I love this far-seeing perspective. Though we should remember that the world of computer science is full of re-invention of old patterns. I would hate to be presented by an obsessive-compulsive MS Office paperclip assistant ;-)

NullOp

This all sounds well and good but I assure you the technology will be put to use in marketing more products on the web. You will not be able to have anything that doesn’t include an ad of some sort. All based upon your browsing history. ‘Cuz your data is really our data don’t ya know!’

Sunny Mishra

wonderful article. gives an insight into what future will bring to our doorsteps.

Bilal Ojjeh ✔

For sure, “Everything is getting smarter”. Question is, will this enable humans to get smarter too? Or will it make them relatively more expensive?

Gagan Saxena

The Cognition-as-a-Service (CaaS) vision is very compelling although the practical implementation aspects have been left at, “…(capabilities) available through API’s”.

The reality is that almost all enterprise systems and most consumer technologies are not set up to consume Cognition or Advanced Analytics Services. Some form of scoring mechanisms are the best cases in use today.

A focus on Operational Decisions is needed first to chalk out the requirements for a Cognition Service. This will help with the design of a Decision Service that can – in turn – plug into the CaaS API’s.

George Abney

My public contribution on single use app personal identity blends profile as encryption key/password; Handshake GPI interface “LetMeWearU” prompt…per LinkedIn.com posting constitutes “Prior Art” serves as introduction to design capabilities. I am available for hire as independent contract consultant, or full-time project R/D participant.

Brandon Wirtz

PlexiNLP is a CaaS.

http://www.stremor.com/

We have API’s already if you want to do all the Semantic like stuff you want against stuff that isn’t in semantic markup. We have API’s if you want to talk to your toaster with voice. We have API’s if you don’t want to read half of the Book you were assigned to do a term paper on.

Kingsley Uyi Idehen

Nova,

As you know the is a world of difference between a concept label and an actual concept. As a moniker of buzz-phrase, “The Semantic Web” might not have resonated with the general Web populace, but as a concept it is thriving beyond all expectations.

As I am sure you know, the concepts that underlie “The Semantic Web” vision are the same concepts that drive:

1. Everything that Google is doing under the “Google Knowledge Graph” banner.

2. Everything Facebook is doing with “Graph Search”.

3. Everything Google, Bing, Yahoo!, and Yandex are doing with regards to Schema.org and HTML hosted structured data islands re., HTML+Microdata, HTML+RDFa, and HTML+JSON=LD.

4. Everything that underlies the Linked Open Data (LOD) — comprised of the likes of DBpedia, Freebase, Wikidata, UK Ordnance Survey, BBC Programmes, Music, Wild Life, Sports, and News.

5. Everything to with “Open Government” as exemplified by the US and UK governments, and many others (even down to the municipality levels).

“Cognition as a Service” is just another label for “Inference and Reasoning” as integral parts of services that are accessible via the World Wide Web. None of that can ever happen of the actual Data on the Web isn’t Web-like and endowed with human- and machine-comprehensible relationship and relation semantics.

Labels come and go, concepts are live forever :-)

Kingsley Uyi Idehen

Revised version of the initial post which is rife with typos as a consequence of touch-typing in haste etc.. It would also help if one could edit posts too :-)

Nova,

As you know there is a world of difference between a concept-label and an actual concept. As a moniker or buzz-phrase, “The Semantic Web” might not have resonated with the general Web populace (end-users or developers), but as a concept it’s thriving beyond all expectations.

As I am sure you know, the concepts that underly “The Semantic Web” vision are the same concepts that drive:

1. Everything that Google is doing under the “Google Knowledge Graph” banner.

2. Everything Facebook is doing with “Graph Search”.

3. Everything Google, Bing, Yahoo!, and Yandex are doing with regards to Schema.org and HTML hosted structured data islands re., HTML+Microdata, HTML+RDFa, and HTML+JSON=LD.

4. Everything that underlies the Linked Open Data (LOD) Cloud — comprised of the likes of DBpedia, Freebase, Wikidata, UK Ordnance Survey, BBC Programmes, Music, Wild Life, Sports, and News.

5. Everything to do with “Open Government” as exemplified by the US and UK governments, and many others (even down to the municipality levels).

“Cognition as a Service” (CaaS) is just another label for “Inference and Reasoning” as integral parts of services that are accessible via the World Wide Web, using HTTP. Fundamentally, CaaS can ever happen if the actual Data on the Web isn’t Web-like and endowed with human- and machine-comprehensible relationship and relation semantics.

In conlcusion, concept-labels come and go, but the fundamental concepts live forever :-)

Kingsley Uyi Idehen

More typo fixing due to missing post edit option:

“Cognition as a Service” (CaaS) is just another label for “Inference and Reasoning” as integral parts of services that are accessible via the World Wide Web, using HTTP. Fundamentally, CaaS *can’t* ever happen if the actual Data on the Web isn’t Web-like and endowed with human- and machine-comprehensible relationship and relation semantics.

In conlcusion, concept-labels come and go, but the fundamental concepts live forever :-)

Marthin Laubscher

Agreed, but beyond that, I also have answers for those impossibly difficult questions involving how to interface with up to 7 billion living brains, so as to use those, with their consent, as one collective intellect of such magnitude that we will without any shadow of a doubt solve the most impossible problems and overcome the most epic of divisions between people. The secrets to this is owned by the people of the world, and I am merely custodian to those. Convince me that you will use it for the benefit only of everybody on earth, and never to help one oppress another, and you may share in these secrets.

Adrian Bowles

You kind of lost credibility with “CaaS will enable every app to become as smart as Siri in its own niche.” I like Siri, I use Siri, but that’s an awfully low bar for real cognitive apps. @sardire’s comment here is spot-on in terms of the current market.

I interviewed 4 of the first Watson ecosystem application partners last month, primarily working in healthcare and retail. Their new apps will be a showcase for the potential of on-demand cognitive computing. Will share the videos on YouTube soon – good insights from people actually building commercial cognitive-based apps today.

@ajbowles

Eray Ozkural

Who knows what new Artificial General Intelligence algorithms can accomplish?

One machine can change the world.

robertsteele

All well and good. but garbage in garbage out. Google for example is completely lacking in integrity and selling search results for money (you get what someone else pays for you to see). Not only are the standard are not there, but the open source everything mindset (see my 2012 paperback by that title) is not there either. INTEGRITY. Such a simple word. Not to be found in the IT industry today.

Derek Litz

I agree completely. Been working for 3 years, and no one sticks up for themselves or good ideals out of fear of dismissal. Comparing a computer to human beings degrades each and every one of us. A computer is not cognitive and it never will be, however, we can make a computer do things in a greater way than any human will be capable of, but the whole point of living is still to engage yourself with other human beings. Otherwise there is no point, and if you compare yourself to a computer you may as well be saying that you are a disposable machine, not much better than a slave. Just like other things we do not understand, computers achieve god-like status from the layman willingly while the people behind the machine are marginalized.

Devices do not get smarter, they become better designed for human beings by human beings.

Steve Ardire

Nova – to be a true CaaS player you need 1) contextual intelligence in the cloud, 2) contextual intelligence at the interface ( remember Tom Gruber’s preso in 2009 at Semantic Technology conference before SIRI was sold to Apple ) and 3) contextual intelligence in chipsets ( you can figure out who the players are here ;-)

Vicarious is still in R&D so LOL and NextIt just uses NLP like many other players in their space and is NOT a real cognitive player ( but you’re an advisor ) and Wolfram got hammered for his hype here Sentient code: An inside look at Stephen Wolfram’s utterly new, insanely ambitious computational paradigm http://goo.gl/5QlKRE

Google of course is a player ( can hit on all 3 criteria above ) but currently there’s only 3 commercial cognitive computing players who offer REAL solutions that can learn and interact naturally with humans to enable human experts to make better decisions in the world of Big Data They are @ibmwatson, @groksolutions ( formerly Numenta ) and @saffrontech.

Sarah

spot on article! Having tried various digital assistants i came across voice Ace chrome extension goo.gl/lOCAF that is as close as it gets to the future digital assistance services described in this post

Comments are closed.