IT operations teams should work with IT service teams, data scientists, and machine learning experts doing the heavy lifting on the build side.

Nathan Eddy, Freelance Writer

May 22, 2023

5 Min Read
AI Assisting Humans and Helping Research Development
sleepyfellow via Alamy Stock

Artificial intelligence can be a game changer when it comes to IT issue resolution, particularly in cloud environments, creating massive productivity gains for organizations. Technology solutions rooted in AI empower teams to monitor and optimize their entire hybrid, multi-cloud topology in real-time.

AI -- especially deterministic or causal AI -- can be used to observe behaviors, search for anomalies and degradations with true business impact, and instantly alert teams when problems occur and point to the root cause.

Andi Grabner, DevOps activist for Dynatrace, says AI and automation can eliminate the need for human intervention to solve basic IT problems. “This frees up time for teams to engage in more complicated issues where their time and touch is more meaningful and offers greater value,” he says.

In this way, AI augments people to perform at a higher level than is otherwise possible.  

“Teams also need to ensure their AI is drawing the correct conclusions, making the right decisions, and implementing the right solutions,” he adds.

Benefits Already Made Clear

Melissa Herrle, vice president of product at OpsRamp, points out AI benefits in IT issue resolution are already being seen today, mainly in reducing alert noise and help desk tickets.

“Intelligent alerting and event correlation ensures that multiple tickets aren’t being filed for the same event,” she explains. “This leads to a reduction in mean time to detect and resolve. This is typically the first benefit customers see from an AIOps implementation.”

Herrle says the smarter the AIOps system becomes, the more the organization can automate processes in response to detected events, be they automated remediation or just automatically directing the right internal resources -- human or machine.

“The biggest limitation is data,” she says. “The more tools in use, the more data has to be collected, filtered and managed, and the more data accuracy issues you’ll have.”

Human feedback can refine and enhance recommendations, then they can be re-purposed into the AIOps system for more accurate recommendations in the future. “That’s certainly one approach to weaving meaningful human touch into AI-based issue resolution tools,” Herrle says. “Along the same lines, I would expect to see more use of generative AI built from internal human knowledge.”

Advantages of NLP and Deep Learning Deployment

Tommy Gardner, CTO of HP Federal, says there are two forms of AI that could impact IT issue resolution: The first is Natural Language Processing (NLP).

“This is starting to be seen through the prototypes of ChatGPT,” he explains. “You can imagine a scenario where you click an IT Support tab or link and speak about your issue rather than chat with someone in IT support.”

The AI algorithm would understand the syntax and content of your issue and search for similar or exact problems and their solutions and report back the most probable solution.

Gardner says the second AI area would be in machine learning (ML) and its subsets in Deep Learning. “Here, the search algorithm of the problem could be trained on the data of past problems in the data lake of issues collected from history or experiences,” he says.

The primary limits of this approach and associated tools are that technology changes over time as new innovations are introduced, improvements are made, and patches are created.

“This implies that solutions or fixes from yesterday, might not apply to today’s product,” Gardner says. “Limits may need to be placed on the length of time you search back in the historical data or limit the database to a specific generation of product.”

If one defines “the human touch” as the instinct that humans have from pattern recognition, the experienced help desk consultant can usually find the shortest path or shortcut to the fix. “The AI may initially provide a comprehensive or step by step process, but the human can at times see beyond the data with their intuition,” he explains.

Responsibilities of the COO, CIO, and CISO

Gardner notes the corporate leadership of a company has the responsibility to optimize corporate operations and ensure that AI is deployed responsibly for various operational needs.

Those leaders are the chief operating officer (COO), the vice president for services, the chief information officer (CIO) and at times the chief information security officer (CISO).

Grabner adds CIOs are often responsible for greenlighting and overseeing AI deployment.

“However, they will work collaboratively with development, IT, security, and business teams to ensure the technology is being implemented cohesively and to optimum effect,” he says.

Gardner notes although generative chatbots like ChatGPT are attracting most of the attention recently, business use of AI for IT-support chatbots has been more commonly adopted by many companies.

“As the cost of data storage goes down and the power of processing goes up, these tools will continue to develop,” he adds. “Competition from other companies will drive the timeline. No one will want to fall behind in this innovation.”

He predicts most will stick their toes in the water and test AI tools out prior to full scale implementation, as companies will want to reduce risk prior to fully jumping in. “Companies will need to set clear goals for implementing new AI tools and assess where non-sensitive opportunities exist and determine ethical guidelines for the company to implement AI, including independent reviews of their AI to avoid mistakes and biases,” Gardner says.

Herrle explains that IT operations teams really want to improve automation -- both automation of IT operations for improving efficiency and productivity and automation of IT processes for improving operational governance.

“You don’t get to automation overnight though,” she notes. “You start with improving topology mapping between applications and infrastructure, consolidating monitoring tools -- especially retiring tools that don’t help you in the cloud -- then intelligent alerting and event correlation. Once you’ve made those leaps, automation of as many operations and processes as possible becomes more realistic.”

What to Read Next:

The Future of AI Is Tiny

5 Ways to Embrace Next-Generation AI

3 Ways AI Can Impact Data Governance

About the Author(s)

Nathan Eddy

Freelance Writer

Nathan Eddy is a freelance writer for InformationWeek. He has written for Popular Mechanics, Sales & Marketing Management Magazine, FierceMarkets, and CRN, among others. In 2012 he made his first documentary film, The Absent Column. He currently lives in Berlin.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights