AI can find IT problems faster, but fixing them is another matter

Recent troubleshooting releases from Infotech and NinjaOne show AI moving deeper into IT ops. Both analyze operational data and recommend next steps without acting independently.

There's a difference between listening to what AI has to say, trusting its recommendations and giving it the autonomy to follow through on its own.

Whether a change is made by an AI agent or a human, someone still has to take responsibility for verifying that it worked -- and be accountable for what happens if it didn't.

Recent research reflects the tension.

A Riverbed survey of 1,200 business and technology respondents found that 90% said their organizations want to pursue autonomous IT ops, while 77% reported reluctance within their organizations to let AI make operational decisions without human approval.

The Riverbed survey doesn't tell us how effective AI is at diagnosing or resolving IT problems. But there's evidence that AI can help IT teams work faster.

In its 2025 State of ITSM report, SolarWinds analyzed service desk data and found that average incident resolution time dropped 17.8% among organizations after they enabled GenAI features. That means incidents took nearly five hours less to resolve on average. The analysis doesn't establish how much of the improvement came directly from AI or whether the underlying problems stayed fixed, though.

More recent SolarWinds research points to another complication. As CIO Dive reported in August, 71% of IT professionals surveyed said their workloads had stayed the same or increased despite the time AI was saving them on individual tasks.

Infotech's CIO Analytics platform includes a service desk module that analyzes and recategorizes tickets to uncover recurring patterns. Its connection to the underlying service desk is read-only, with human analysts helping CIOs interpret recommendations.

NinjaOne's Katana AI assistant examines endpoint information, identifies possible causes and recommends troubleshooting actions, including scripts. The product is in early access and, according to NinjaOne documentation, requires technician approval before suggested actions run.

Together, the products raise a familiar IT question: What happens after AI identifies a problem?

Finding the problem is only the beginning

Support tickets can reveal where enterprise software is getting in the way of work. Recurring problems with reports, workflows or permissions often show up there before anyone realizes something larger needs attention.

Help desk teams need to look beyond how quickly tickets are handled. Repeated complaints about an application or workflow deserve a closer look.

AI can now analyze more operational information, recognize patterns and recommend actions without requiring technicians to piece together every connection manually.

The harder part is figuring out what's causing the problem.

Consider repeated access requests. A technician might grant the missing permission each time, and an AI assistant might recommend doing the same thing more quickly.

The cause could be an outdated role, an approval process that no longer reflects how employees work or a security restriction that should remain in place.

Who gets to authorize the change?

Ticket analysis needs business context. The support team knows which questions keep coming up. The process owner understands why employees are running into trouble. The application owner can look at the settings, permissions or recent changes that might be involved.

IT may control the setting, but the business owner needs to help decide what it should allow. That distinction matters when AI progresses from suggesting a fix to executing one.

An approved remediation script might address a known endpoint problem. Changing employee permissions or modifying an approval workflow could have consequences that extend beyond IT support.

CIOs need to know which actions can be safely automated, which require human approval and when the decision belongs to someone outside IT altogether.

Human approval helps establish those boundaries, especially when a change could affect employees, business processes or other systems.

Trust in AI-driven IT automation can vary by task. In this IT Ops Query video from Dell Technologies World, HyperFrame Research CEO and principal analyst Steven Dickens discusses how enterprise buyers can evaluate AI automation tools and determine where human oversight remains necessary.

A completed action isn't necessarily a solved problem

Some organizations already demonstrate how effective automation can be when the work is clearly defined.

 At A+E Global Media, technology executive Chris Kairinos told Computer Weekly that the company automated more than 30 service-catalog items, accounting for approximately 5,000 requests annually. Average resolution time for those routine requests reportedly fell from about 17 hours to minutes.

A script runs successfully, a technician approves a change and the ticket closes. The same problem could still return.

That's a significant operational improvement involving well-defined tasks. Recurring problems with uncertain causes can require considerably more investigation.

A script runs successfully, a technician approves a change and the ticket closes. The same problem could still return. If it does, IT needs to figure out why.

That includes verifying the result. Did the access problem stop recurring? Did employees complete the workflow without another workaround? Did the change create an exception somewhere else?

Reviewing recurring tickets can show IT leaders where enterprise systems or workflows need attention. AI-assisted IT ops can make those patterns easier to see. The people responsible for the affected work must decide what should change and monitor the results.

Someone still has to establish whether the fix worked.

James Alan Miller is a veteran technology editor and writer and Lead Editor for CIO News at Informa TechTarget. He directs coverage of enterprise technology strategy, AI, software, data, infrastructure and the decisions shaping how CIOs manage increasingly complex IT environments.