Does a help desk pro need to know AI?

As employers look for AI talent in their data management and IT ranks, what’s a poor ol’ password-resetting, printer-kicking help desk professional to do? Does a tech support pro need to become an AI master, too?

The short answer, according to IT practitioners who spoke with us: Yes.

AI-powered automation has hit the help desk, and the help desk pros, in turn, need to know what’s available, what’s possible, and what’s not.

A specific set of skills. According to online education platform edX’s August review of job postings, over 120,000 unique listings featured a need for AI skills—a 104% uptick year over year. Top AI-related skills, edX found, included machine learning, deep learning, and large language modeling—not exactly easy to learn overnight.

According to a November study by hiring site Upwork, companies are increasingly seeking data mining and management skills; requests for such expertise grew 26% month over month in October. AI and machine learning skills—including data annotation, labeling, and generative AI modeling—also grew 15% month to month. A recent poll from nonprofit tech org IEEE found “ethical practices skills” as a top expertise sought by tech pros.

So, what does that mean for the help desk professionals?

CompTIA’s Robert Rohrman, SVP of IT infrastructure, and his team may not be battling deep, life and death questions related to ethics and the implications of AI use. They may just need a quick answer to a troubleshooting question, like what a certain error number means.

“I think every worker can benefit from a familiarity and comfort level with various AI tools on the help desk,” Rohrman said.

But IT teams also face more complex AI implementations. Though a recent survey from consulting firm McKinsey & Company concluded that orgs overall are still in an experimentation and pilot stage with their “use of AI,” IT is the business unit most often “scaling” agents these days. (The report had 8% of respondents reporting IT’s scaling of agentic AI—with 13% experimenting or piloting.)

Agents are hitting the help desk. Unthread—a Slack- and Teams-based support tool—uses agentic capabilities to find common issues and turn conversational chat messages into automatic tickets and remediation where possible. Meanwhile, companies like Simbian offer an AI platform for security operations centers (SOCs), which automatically responds to alerts.

As automation takes over tasks previously done by SOC analysts, Sumedh Barde, Simbian’s chief product officer, sees security pros assuming a broader role: the generalist analyst. Supported by an AI handling tasks like vulnerability management, this generalist will be able to focus on business-context questions like, “What is the worst-case business impact if an asset is compromised?”

The wider-ranging generalist role, he thinks, will certainly handle an agent’s inevitable errors—like fixes related to permissions, incorrect data, or faulty outputs related to that incorrect data.

Unthread’s CEO Tom Bachant thinks help desk pros need to learn Model Context Protocol. The “MCP” connector allows easy, standard integration between a large language model and an external but accessible data source. An IT pro, for example, may need to provision an MCP server (with minimal access for security purposes, Bachant suggested) as the org pulls from a third-party software tool for insights.

Market intelligence firm Gartner predicts that, by 2029, agentic AI “will autonomously resolve 80% of common customer service issues without human intervention.”

Know your limits. Part of an IT practitioner’s skill set, according to Yoav Susz, GM for US at Atera, will be knowing all the different AI-powered ways to make effortful tasks more effortless—for example, a capability that summarizes the full history of a ticket so the help desk pro doesn’t have to read a full timeline.

Just as important as knowing what AI can do for you: knowing what it cannot. James Stanger, chief technology evangelist at CompTIA, thinks a good tech support pro will use AI with caution, considering the possibility of imperfect outputs.

“They need to have a healthy skepticism about the results,” Stanger said. “And a lot of help desk professionals who are just starting may not have that healthy skepticism, so they might bring in some of those tools that will cause more problems than solve.”