IBM has traded widely on the prowess of its cadre of supercomputers. Built for speed and sheer computing capacity, IBM Blue Gene has a petaflops pace almost unfathomable to the human brain; and then there is Watson, the natural language supercomputer with Artificial Intelligence named after IBM’s first CEO.
Watson has proved extremely good at consuming, analysing and making decisions based upon the unstructured content with which it is fed. So good in fact that the computer beat two human contestant ‘previous champions’ in a quiz shoot-out on the US game show Jeopardy.
But Watson was destined for more. IBM decided that the machine-learning technology behind this box of processors and chips could be put to greater use for humanity.
Watson’s “intelligence” has since been directed at worthy objectives in health, particularly in making prognoses - where it's thought to be able to out-think even the most knowledgeable medical minds. Other applications could see the legal trade being usurped by a new kind of lawyer, ie one with a computerised knowledge of all statute books and points of precedent, capable of deduction based upon incomprehensibly huge streams of historical (and, conceivably, real-time) data.
What next then? Well it’s going to be that word that all vendors just love to use to denote their technology’s breadth - and that term is “platform”.
Anything-as-a-Platform is the term sitting somewhere in between being a noun and a verb that describes the point at which a database vendor, cloud hosting supplier, technology services company etc. decides that software application developers should be able to “program to” their technology proposition.
Watson is becoming a platform rather than just a very impressive box, by dint of IBM readying itself to give external third party developers access to its power. The company is doing this by opening up an API set , building what it will eventually call a Watson app marketplace.
If successful, the Watson “app store” could be a route for software programmers to eventually monetise their own innovative “applications”, on the back of the AI backbone that they tap into.
Will it be successful? Well, let’s be honest, yes it will. IBM doesn’t fail at this kind of thing and the firm is using a RESTful API set and familiar programming tools to maximise the chances of user take-up and adoption. It fits the global IBM model nicely too as the firm could subsequently put itself in a position to sell integration services on top (remember IBM Global Services?) and even international translation perhaps.
If Google is selling routes and channels and formats to access its data, IBM is selling access to its data processing and analysis, so what’s wrong with that?
If there is a criticism to be levied here it comes back to the term platform again. Where we see developers urged to “program to” Watson, the more purist of mind will still regard a platform as something to “program on” or even “with” at the very least.
In other words, a platform is something with an accompanying language, plus a set of tools, components and a runtime: think .NET or think Java. OK so fair enough, this is not the only way to think of a platform now and the definition is widening, but it is being widened by commercial labelling rather than from technology functionality in the first instance.
Let’s also remember that this is a cloud computing practitioner’s website and highlight the fact that Watson (as a platform) will now also be offered as a service.
As a piece of terminology, computational artificial intelligence & knowledge engine platform -as-a-Service (or CAIKEP-aaS) isn’t going to stick. So you will hear IBM talk about Watson as a service in the same way that Google and perhaps also Wolfram Research will position Wolfram Alpha.
So is there one problem that IBM Watson can’t solve? One could suggest that it needs to shrug off as much of the ‘old traditional IBM’ feel if it is to be perceived as cutting edge and worth programmers’ attentions. But perhaps this heritage for big machines is exactly the kind of image that it should present? At least Watson can still occasionally get some things wrong and doesn’t know where Chicago is. There is hope for us humans yes, just.
