Artificial intelligence (AI) is a hot topic in the enterprise these days, with industry leaders looking at applications ranging from smart products to self-healing – even self-aware – computing infrastructure.

But how much of this is real and how much is science fiction? Are we really on the verge of selling out our humanity to a class of robot overlords? Or will the technology fail to produce any meaningful change at all?

Judging by what is available right now and where the development trends are heading, the answer to the last two questions is "No."

AI vs Automation

The first thing to understand about today’s AI is that it is not just an extension of existing automation. Traditional automation can be used to make machines, devices and applications perform repeatable tasks, usually at a consistent rate and in a consistent manner. AI-driven automation allows the programmed entity to first adapt and respond to a wide range of stimuli and then adjust its own programming and operating patterns to suit its changing environment. So while an automated robotic arm can be programmed to attach a certain panel to a certain kind of car door the same way an infinite number of times, an AI arm can analyze different kinds of panels and figure out on its own how to attach them to different kinds of doors. (To learn more about automation, see Automation: The Future of Data Science and Machine Learning?)

In terms of enterprise infrastructure, AI is the key to implementing the digital transformation that is necessary to thrive in a service-oriented economy, says Venkat Srinivasan, chairman and CEO of automation firm Rage Frameworks. AI is already introducing several key capabilities to infrastructure operations using a more linguistics approach to data analysis in place of traditional database algorithms. In this way, enterprise data systems gain the ability to understand data in its context and relevance to the real world, which in turn allows them to make sense of the reams of unstructured data that is sitting in enterprise archives untouched and forgotten. At the same time, it enables a higher level of reasoning and traceability, giving human operators and other intelligent systems the ability to drill down into analytics and other processes to determine how and why decisions are being made.

But how, exactly, will this all play out on an operational level? What kinds of applications can we expect to see from AI-driven processes?

According to Gil Press, managing partner at research consultancy gPress, two of the more profound are speech recognition and natural language generation. Using neural networks and other advanced technologies, companies like Google and Amazon are already pushing conversational computing into the home through Google Home and Alexa. It’s only a matter of time, then, that these same technologies invade the data center, allowing even non-technical users to simply ask their data environments what they need to know rather than typing, clicking or texting. As well, with the self-learning, self-correcting capabilities that AI brings to the table, it is likely that systems life cycle and upgrade patterns will change dramatically – equipment won’t degrade over time; it will get better with little or no human involvement. As well, the data environment itself will become more proactive in its operations, making suggestions as to how to optimize data performance, not just responding to commands.

Any Downsides?

Is this vision of a bright, shiny future all there is to AI in the enterprise then? What about the downsides?

To be sure, says eWeek’s Chris Preimesberger, AI will have to be implemented in a controlled, coordinated fashion, just like any other technology. In fact, many of the key pitfalls are the same as with existing data platforms, such as deploying a technology in search of a solution and failing to ensure automated processes are germane to business requirements. But AI also requires some special attention, such as recognizing the fact that AI can only deliver results that are as good as the data it receives. There is also a trade-off between breadth and depth when it comes to AI; any system that is designed to address a broad range of functions will not be able to drill down into the highly granular processes that drive productivity. (For more on the future of AI, see Don’t Look Back, Here They Come! The Advance of Artificial Intelligence.)

And probably most important of all: no matter how “smart” an AI platform becomes, it will always need a human brain to guide it.

So although it may sound cliché, the fact is that AI is truly on the verge or remaking the data environment to something akin to what we’ve seen in sci-fi movies all these years: a talking, thinking data environment that is literally all around us, like the onboard computer of the Starship Enterprise.

In this light, it seems we will all have to get used to the idea that the enterprise is no longer just a collection of devices and software that supports our data, but a responsive and highly effective member of the business team.