It was hard to see IBM's Watson supercomputer as much more than a PR stunt. Even if the machine can win at "Jeopardy," that doesn't necessarily make it useful outside of such a closed-ended experiment.
But IBM must have a lot of confidence in its newly revamped Watson machine-learning technology, because it's preparing to make Watson into a cloud service by way of a service and an API set.
Watson 2.0, as it might be called, now uses a dramatically smaller hardware array (16 to 32 cores instead of the original Watson's 2,900), but it runs three times faster than its predecessor. It's initially being offered to only a small, select group of IBM partners, which means IBM wants to build ties first and foremost with outfits that can program for Watson in a robust way. But even the sketchy hints that have leaked so far are revealing.
One of the details about plans for the Watson cloud service is an app marketplace. If IBM follows the models set out by the likes of everyone from Apple to HP, it could become a way for companies to monetize not just access to a specific data set, but a custom way of processing that data set. For example, machine translation: It ought to be possible for IBM to allow a company to sell apps for translating to or from specific languages, with each language pair being its own app and a discount for packs of multiple languages.
Another key issue will be IBM's ability to attract programmers to the platform. By using a RESTful API set and other familiar programming tools, IBM evidently wants to keep the bar for entry low enough to be useful. Since Watson learns from all interactions that take place within its cloud, the broader the variety of applications, the better.
A third and so far unexplored issue is what could be called "emergent intellectual property." As the New York Times pointed out in its coverage of the new Watson, IBM retains the rights to certain improvements for Watson that emerge because of customer interactions (such as being able to interact with people in a new language).
Fourth is how IBM plans to monetize this. So far it has offered no clues as to how it might do this, whether by a kind of internal, abstract metric à la Amazon EC2's "compute units" or by simpler (if more regressive) real-world units like amount of storage used or API requests per time unit. The sheer newness of what IBM is attempting to do may force it to adopt a metric that simply doesn't parallel anything else out there.
IBM's likely to be in competition with at least two other purveyors of intelligent systems. One is Wolfram Research, whose Wolfram Alpha "computational knowledge engine" has spawned divisive opinions about its true usefulness. It supplies back-end knowledge for Apple's Siri, but its most newsworthy front-end achievement of late was to add every single Pokémon to its database.
The biggest elephant in the room is Google, of course. What gives Google a possible edge is its unprecedented access to more raw information -- and more kinds of it -- than just about anyone else. What's less clear is if it has the chops to deliver something truly useful and battle-tested to some degree. If in fact Google has something under wraps right now to rival both Wolfram and IBM, it would be high time to start talking about it.
This story, "Watson as a service: IBM preps AI in the cloud," was originally published at InfoWorld.com. Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest developments in business technology news, follow InfoWorld.com on Twitter.