Smart machines: separating fact from fiction

Gartner predicts that smart machines will be a top five investment priority for more than 30 per cent of CIOs by 2020. Tom Austin of Gartner writes why.

Comments

It’s important to exercise extreme caution when exploring investments predicated on assuming that machines are really smart and that AI has finally succeeded in producing tools that can emulate how the human brain works or replicate our cognitive processes.

We’re at the start of a smart machine age, precipitated by a combination of three developments – radical new hardware, massive amounts of data and unprecedented advances in deep neural networks - that came together around 2012. The journey will be as transformative (and disruptive) as travelling through the industrial revolution.

Physical things large and small will get smarter, becoming more capable and able to understand and respond to our needs. They will also become safer and more adaptable. Think of cars, retail stores, houses, offices, factories and entertainment venues — virtually all forms of intellectual work and social interaction will change significantly as machines help us excel.

Smart machines will affect every job, career and industry in ways we only dimly perceive today.

There are real business opportunities that are already allowing enterprises to achieve very positive results, such as autonomous vehicles, smart vision systems, virtual customer assistants, smart (personal) agents and natural language processing.

As a CIO, you don’t have to understand all of the capabilities and limits. However, you do need to ensure your cross-functional teams are well enough versed to be able to exploit the opportunities and disabuse the myths, both of which continue to emerge in this space.

Tom Austin, Gartner

Cut through the hype

Marketers (and popular media) confuse fact and fiction when implying that smart machines have humanlike capabilities. It’s important to separate the hype from reality.

Assigning human attributes to technology (anthropomorphism) distorts our understanding of what that technology can truly accomplish. This is why Gartner prefers the generic, umbrella term ‘smart machines’ over terms like cognitive computing and artificial intelligence.

When we ascribe human characteristics to technologies we run three direct risks:

Deceiving ourselves about the real capabilities and limitations of the technology;
Overgeneralising the human related attributes to the technology; and
Constructing, or allowing others to construct, scary, non-productive fictional tales about the technology.

The consequences inherent in these risks include disappointment, wasted time and money, negative career and business impacts, calls for regulation and government intervention, and a loss of momentum for very promising technologies.

What’s the difference?

To cut through the fog (and resulting foggy decision making), let’s explore the difference between the terms smart machines, cognitive and AI.

1) Smart machines – These technologies adapt their behaviour based on experience, are not totally dependent on instructions from people (they learn on their own) and are able to come up with unanticipated results. However, smart machines are "smart" in a narrow sense. These technologies are not generally able to reason on their own, demonstrate common sense or figure out new ways of doing things.

2) AI – Back in 1955, researchers assumed they could describe the processes that make up human intelligence and automate them, creating an artificial (human) intelligence. I believe they were wrong then and remain wrong now, but we are moving beyond that into the smart machine age.

3) Cognitive computing – This term leads people to assign human cognitive properties to computing that are not evident in technologies. Machines are not cognate – they don't think and they don't reason. Their "reasoning" typically reflects the use of reasoning rules (detailed algorithms) coded for them to run and the data that has been used to train them. Machines have no common sense and they are narrow in purpose. There’s nothing in research literature to suggest otherwise, at least not yet.

Real business opportunities

Despite the hype and confusion, there are many practical applications of smart machine technologies that can do (at scale) what required expensive human labour or could not be done at all five years ago. For example:

In the US, Go Moment's Ivy is a virtual customer assistant for hotel guests that addresses guests' questions and dispatches requests to the right point of contact for improved efficiency and guest delight.
Canada’s Yactraq adds value for organisations by applying natural-language processing to business applications. It excels at the quick, inexpensive, automated creation of custom vocabularies that make it possible to extract more insights out of audio information.
US-based Clarifai represents practical bleeding-edge technology, delivered as a service for image and video classification, tagging and search, as well as for value-added applications that target specific tasks, such as food classification, website and social media analysis and website moderation.

Smart machine capabilities are now sufficiently stable and transparent that organisations can experiment with them to create business value. Business investment in smart machines will grow quickly, focusing on revenue growth and operational efficiency. Gartner predicts that smart machines will be a top five investment priority for more than 30 per cent of CIOs by 2020.

Read more: Thinking ahead with AI

Tom Austin is a vice president at Gartner and has been a Gartner Fellow since 1997. He drives Gartner's research content incubator (the Maverick Program) and leads a research community on the "smart machine" age. He is also Gartner’s lead analyst on Google.