d1revolver - Fotolia
ORLANDO, Fla. -- Imagine a world where you need transportation to an upcoming meeting, and your personal digital assistant autonomously requests a self-driving car to meet you at the exact time that you need it. That's the basic premise of "smart machines" in a session presented by Tom Austin, Gartner vice president and fellow at Gartner ITxpo 2014. Austin said that smart machines will use a mix of content and context to open up a new era in human-machine collaboration. Whether the notion excites or frightens you, it's time to take a closer look at what smart machines are and how they can impact businesses in the years ahead.
Defining a smart machine
We've long imagined a collaborative relationship with intelligent machines. In 1968, the science fiction film 2001: A Space Odyssey featured the character HAL 2000, an artificial intelligence system capable of interacting with its human crew and independently operating an interplanetary space flight. In the 1986 classic Alien, the character Bishop was a fully realized android. Today, we have Apple's Siri and Microsoft's Cortana pushing the limitations of speech recognition and interactive assistance on our smartphones, along with learning robots that can do simple household chores. They're not able to pilot a spacecraft yet, but Austin said that day might be not too far away.
So what is a smart machine, anyway? Austin explained seven principal characteristics that go into any smart machine. First, they deal with complexity (far more complexity than a human can realistically handle). They are probabilistic predictors designed to anticipate future activity based on previous activity. They actively learn by querying you for input and passively learn by watching your actions. They can act autonomously to some extent. They appear to have human-like understanding. And they have a particular purpose.
Real products and services are out there now. One CIO from a global recovery audit firm reported using smart machines to establish correlations across complex data sets. "We do a lot of algorithmic-type coding," he said. "Our system can start making interpretive diagnostics."
Smart machines, as we see them appearing today, arise from a convergence of four important areas: an explosion of digital content and media; the development of algorithms capable of selecting, ingesting and processing that content; the continued evolution of computing hardware capable of effectively running such algorithms; and the availability of a global network like the Internet that ties everything together.
Austin broke down smart machines into physical and virtual categories. Physical smart machines include autonomous vehicles like GM's almost self-driving Cadillac and learning robots like Baxter that are currently being developed. Virtual smart machines include image recognition, speech recognition and language translation platforms along with all types of intelligent personal assistants (such as Siri and Cortana).
All of these smart machines are in very early stages of development, and have a long road before reaching their full potential. At the same time, not every organization is prepared to implement such technologies on a major scale. An enterprise architect for a major fashion retailer attending the session wasn't so sure. "Not all of the organization is ready [for smart machines], but some segments are."
Preparing for smart machines
The potential power of smart machines can elicit real fear among some users. After all, in 2001: A Space Odyssey, HAL 2000 did wind up killing all aboard the Discovery spacecraft except Dr. David Bowman, who had to pull the plug on the rogue machine. But Austin emphasizes a pragmatic view of the risks, noting a potential for employee displacement, legal liability and the effects of questionable integrity or reliability of the underlying content (harkening back to the old computing adage, garbage in, garbage out).
But perhaps more relevant concerns surround issues of privacy and trust. At some point, users must understand how much of their activity the smart machine can track and report. Users must also be able to exercise a level of control over how much autonomy the machine can shoulder. For example, maybe it's OK for the virtual private assistant to remind you about upcoming appointments, but you might not want the machine booking flights or calling cabs for you.
Companies interested in using smart machines shouldn't rely on a single vendor. The smart machine market has not yet solidified into a single stack provider. "A single-vendor strategy creates major strategic disadvantages," Austin said. "No one vendor is the master of all [smart machines]." Many new small vendors are also carving noteworthy niches in the market, from major vendors like Google and Microsoft, to relatively small players like Interactions Corp.'s virtual assistant, IPsoft's autonomic and cognitive platform and x.ai's virtual assistant, among others.
Ultimately, Austin says that smart machine technology represents a powerful new way for humans and machines to interact -- offering unimagined possibilities for the business. "Laggards lose," he said, noting that competitors that choose to embrace the technology may quickly outmaneuver you and change the nature of your industry while you're not even looking. He expects smart machine technologies to ramp up significantly in 2017, and suggests that organizations start exploring the technology into 2015 if possible.
Using smart machine technology for real-time supply chain data
Understanding the Internet of Things market
The interoperability challenge of the industrial Internet