The term AI (Artificial Intelligence) is being thrown around left and right these days. Many companies claim they have an AI play even when they don’t. But there’s another type of AI—an algorithmic approach to intelligence—that is smart and is emerging as the type of AI that IT organizations of all types could start implementing very soon. And there’s nothing artificial about it.

AIOps stands for Algorithmic IT Operations and is a new category as defined by Gartner research that is an evolution of what the industry previously referred to as ITOA (IT Operations and Analytics). We have reached a point where data science and algorithms are being successfully applied to automate traditionally manual tasks and processes in IT Operations. Now, algorithmics are being incorporated into tools that allow organizations to streamline operations even further by liberating humans from time-consuming and error prone processes, such as defining and managing an endless sprawl of rules and filters in legacy IT Management systems.

The technologies introduced to the market over the last several years, utilizing data science and machine learning in response to the increasing complexities of enterprises undergoing digital transformations, have led to the birth of the term ‘AIOps’ (Algorithmic IT Operations). Gartner reports that by 2020, approximately 50% of enterprises will be actively using AIOps platforms to provide insight into both business execution and IT Operations, which is an increase from fewer than 10% today.

To understand the difference between AIOps, and generic AI, we need to look at AI in more detail.

A Short History of AI

AI is a term we use to describe machines (or software) that mimic the human cognitive processes. That is, the ability to think in the same way a human does. AI got a huge boost in the 1940s with Alan Turing’s work, but it has only really come of age very recently, as the economics and power of computing resources have always been a barrier.

The question is, other than the whimsical science fiction fueled notions, why would we want machines to mimic humans? The answer to that also explains why certain applications of AI are proving more successful than others. AI’s strength is in solving human-oriented problems, so we see solutions being applied to products requiring complex marketing interactions, self-driving cars, behavioral analysis and so forth.

IT Operations however is a slightly different environment. In many cases you are not directly managing humans like you would with service desks, for example. With IT Operations, you are interacting with applications, software and infrastructure. While they can be incredibly complex and unpredictable, they are not human.

Human Thinking v. Machine Thinking

This is where the subtle difference of Algorithmic IT Operations comes in. Solutions in this space are going to be very focused on problem solving, and in many cases using algorithmic techniques that closely emulate what humans would do (but at a far greater speed and scale). The difference is that the learning processes will acknowledge the machine nature of the subject. The efficiency of algorithms (within a much more limited scope of “thinking”) is what drives up the value of AIOps, versus human intelligence – the kind that is almost limitless in scope, but less efficient than a computer. 

Of course, humans are essential to effective IT Operations too. A goal of AIOps solutions is to make life better for us, but the line gets a bit blurry when humans interact with AIOps. The more advanced AIOps solutions will have neural-network technology built in that will learn from its operators, adapt and attempt to eliminate repetitive and tedious tasks.

The Companies of the Future

Why do companies need AIOps? Modern IT environments are incredibly (and increasingly) complex and ever-changing, leading to large amounts of time and resources devoted to monitoring, troubleshooting and course-correcting. It’s a reactive position for most companies, but when teams use AIOps technology they can leverage change-tolerant algorithms and access indexed information. This allows them to spend more time focused on proactive, meaningful work rather than fixing the same problems repeatedly or spending time managing rules and filters.

What do we mean when we say “rules”? Rules—the kind of logic that can be described in “if this, then do that” statements—are very effective for simple use-cases, but scale very poorly. On the other hand, the use of algorithms and associated machine learning techniques provides more flexible expressions that are not only more powerful, but also less “brittle” and can accommodate the inevitable changes that occur over time. This translates to more effective solutions and lower cost of ownership. The challenge for vendors is to provide solutions that package the technology such that the user doesn’t have to be exposed to the underlying complexity. It’s not sufficient to provide a tool-kit, as that just means a business has to start hiring data science professionals instead of engineers.

Going Forward

With the technological advancements of algorithmic intelligence, what typically takes humans hours to accomplish can now be done in just seconds in an automated manner and with much better results. Traditional IT Ops teams manage high alert volumes, leading to major distractions and time spent doing mundane work instead of innovating. These problems are solved when IT teams start using AIOps to supplement operators as they sort through seas of alerts and noise. The benefits of AIOps adoption are being recognized by enterprises who are rapidly on-boarding these technologies to improve their customers’ digital experience across industries – including banking, entertainment, transportation, retail and even government.

While AIOps is a relatively new term, this does not mean that it’s a future trend to be evaluated. In this digital age, any organization using traditional techniques to manage machine data is either ignoring valuable information, or overloading its already strained operations teams. With the explosion in data sources, it’s imperative for CIOs to rapidly embrace AIOps technologies in order to establish a data-centric approach. While traditional AI will still offer unique benefits in some areas, it’s the problem-solving focus of AIOps that will provide the most immediate and profound value that businesses seek.

Like this article? Subscribe to our weekly newsletter to never miss out!

Image: x6e38CC BY 2.0

Previous post

White Paper: The Importance of a Strategic Response to Cyber Incidents

Next post

Three Key Facts About Sensors That Are Driving IoT Forward