AI advantages in ITSM: The role and impact of AI in ITSM

With all the advances in the area of Artificial Intelligence (AI) and its widespread application across various disciplines, this new technology is making its way to IT Service Management (ITSM). ITSM has seen multiple waves of new technology, each promising to redefine the way things work. But many of them made little to no impression and have passed on as mere fads.

Now, the obvious question on everyone’s mind is: Will AI actually make ITSM easier and more efficient? That’s the question we’ll address in this two-part series, “The AI Advantage in ITSM.” In part one, “AI at Work in ITSM,” we’ll set the context for our AI discussion. In part two, “Features and Use Cases,” we’ll look at specific AI-based features and use case scenarios poised to change the way IT service desks work.

Industry experts have some strong predictions about this. Gartner states, in its Predicts 2018: Artificial Intelligence report[i], that by 2022, 40 percent of customer-facing employees and citizen-facing government workers will consult daily an AI virtual support agent for decision or process support. Gartner adds that AI capabilities will power virtual support agents as a resource enabling human support agents to respond faster and more efficiently to customer/citizen inquires or actions.

AI will start having a real impact on our IT service desks once it can perform actions that humans are bad at and perform actions that humans would rather not do. These actions can fall into one of three categories: smart automations, strategic insights, and predictive analytics.

For example, routing incoming tickets manually consumes a lot of time — time an IT technician could use for more important tasks. Some help desks have automated ticket routing by defining rules that categorize requests based on preset conditions and parameters, but these rules are static, meaning they won’t adapt or improve with time.

With the help of AI technology like Machine Learning (ML), service desks can create a categorization model based on historic IT service desk data. Best of all, these ML models will become more accurate over time by taking live data into account. Such ML-based models are more efficient than manual categorization or rule-based automations.

Vendors can create similar AI-based models to generate insights and predict anomalies in IT service desks, which otherwise takes a lot of time, effort and skill from humans. Some real-life scenarios might include suggesting the right window for patch updates, aiding in change planning and implementation, flagging requests that could violate an SLA, and predicting IT problems.

Initial applications of AI in ITSM

The next question everyone wants an answer to is: Where will it all start?

Even with the expansion of AI applications across various fields, the AI technology with the furthest reach today is virtual assistants. Almost every smartphone today has a virtual assistant that helps people with shopping, travel, finance management, time management, and more. And with technologies like Google Duplex taking shape, the role of these virtual assistants is bound to expand soon.

Similarly, the first application of AI in IT service desks will likely be chatbots and virtual IT service assistants. Though not immediately, AI-based virtual assistants will probably replace humans to become the first point of contact between end users and the IT service desk. We will get a small preview of this with context-specific chatbots that can help take some of the load off technicians by handling simple requests.

For example, there are specific chatbots that helps us with any questions related to the GDPR or security and privacy issues. We also have a chatbot, rightly named Jeeves, that updates our lunch and dinner menus. These kinds of chatbots operate with a knowledge base as their foundation. If these bots reach a point where they run out of options or answers, they pull in a human to finish the job. Chatbots can also help users perform other simple IT service desk operations like creating a ticket, raising an asset request, or even requesting a password reset.

Below are some simple AI-based features that will make their way into IT service desks:

• Automatic categorization of incidents
• Intelligent agent assignment for incoming requests
• Anomaly detection by flagging unusual repeat incidents
• Using predictive analytics to flag requests that could violate SLAs

How AI works in ITSM

AI algorithms and applications are developed based on the available documented knowledge and historic data; that means AI is as effective as the knowledge base and data it’s developed on. Similarly, in ITSM, to develop an AI-based model for any specific context, there has to be a properly documented set of resolutions, workarounds, knowledge articles, and well-maintained historical data. For example, to train an AI-based categorization or prioritization model, we need a historic database of all requests with parameters such as request type, level, impact, urgency and site, and it all needs to be properly documented.

On top of everything, AI-based models like these aren’t universal, which means while a certain model may work for one service desk, it likely won’t work for many others. Categorization and prioritization models are trained on a specific data set and only work for the service desk from which that data set is pulled. These models continuously train themselves with live data to increase their accuracy and effectiveness over time.

With the AI groundwork laid above, please return for part two of “The AI Advantage in ITSM” to learn about specific, AI-based features and use cases.

This article was originally published in DATAVERSITY.