How to lead through AI anxiety without stalling progress

From boardroom pressure to employee fear, AI anxiety isn’t a technology problem

Image:
Herman Brown, CIO, Office of the District Attorney, City and County of San Francisco, speaks at an Executive Session at the MES Spring Summit in Houston. Image: Samara Lynn/The Channel Company

At MES Spring in Houston this week, one CIO explained why AI anxiety is a leadership issue, and how to manage it without rushing AI into failure.

Workforce anxiety around AI is real, and everyone’s feeling it.

C-suite executives, investors, and other stakeholders are pushing AI use in a frenetic state of FOMO — eager to reap the ROI and streamlined business processes AI promises. Workers are urged to adopt AI quickly, while having to quietly manage their fears of becoming obsolete.

Turning AI pilots into successful production outcomes relies on all parties in an organisation sharing the same realistic expectations about what AI can and cannot do.

Failed AI initiatives often begin when CIOs misdiagnose AI anxiety as a technology or HR issue rather than a leadership challenge. It is not. AI anxiety is a leadership issue, and IT leaders must proactively address mistrust and other concerns about AI, advises one CIO.

During an executive session at the MES Midsize Enterprise Summit Spring 2026 in Houston, Herman Brown, CIO at the district attorney’s office in San Francisco, offered advice to attendees on taking a leadership role in quelling AI anxiety in their organisations.

[RELATED: IT leaders buying AI need an RFP, not a pitch deck]

First and foremost, CIOs should remain steadfast against rushed AI plans.

Brown recalled an incident where an email was sent to 32,000 San Francisco employees urging them to “go out and start using AI.”

“I had to go to my leadership team and [I] said, ‘Absolutely not.’ I called the city CIO and said you can’t [send] my staff … these types of emails. You want to send something it needs to come through me.”

Organisations are working through establishing the internal company pecking order for AI accountability. Brown took the reins and positioned himself as an AI leader, not waiting for delegation. His next move was to set up a solid AI policy.

He also shared other actions for IT leaders to take ownership and address concerns about AI:

Demonstrate how AI replaces labour, not laborers

AI’s real impact is on task-level work, not entire roles. Historical examples - from ATMs to mobile banking - show that automation removes routine work but creates demand for higher-skill roles, new functions and more oversight.

Expect higher demand for mid-level and senior talent who can supervise, integrate and validate AI-driven output.

Cutting junior roles too aggressively can backfire when AI output outpaces your team’s ability to manage it.

Instead, actively reset the narrative: AI is here to augment productivity, not eliminate motivated workers.

“It’s natural for people to have that anxiety about AI, because there are mindsets out there that AI is going to replace people.”

Clarity beats reassurance, said Brown. Saying “AI won’t replace you” isn’t enough; leaders must show how roles evolve.

Prevent ‘Just try’ AI without guardrails

Letting employees “just try AI” without guardrails is dangerous, especially for organisations handling regulated or sensitive data.

“Just because we can do something, does that mean we should do it?” Brown asked.

“We have an ethical responsibility – to the community, to our organisation, to ourselves. … Governance is about putting guardrails in place, not stopping innovation,” he added.

AI policies must be local, contextual and enforceable, not broad corporate directives.

Ethical responsibility matters even where regulations lag (e.g., sensitive or human-impact data).

Governance is not about blocking innovation, rather, it’s about controlling risk and accountability.

Create an AI strategy that starts with task mapping, not tools

Brown emphasised the need to establish a practical framework. Start by listing your weekly tasks and sorting them into three buckets:

  1. AI can fully automate
  2. AI assists, but a human stays in the loop
  3. AI cannot do this (human-only work)

“Bucket one: AI can do it for me. Bucket two: AI can help, but a human stays in the middle. Bucket three: things AI can’t do,” is how he outlined the framework.

ROI comes fastest from bucket-one tasks (basic reporting, summaries, repetitive workflows); but competitive advantage comes from investing in bucket-three skills.

Don’t over-automate human interaction

Using AI to over-automate traditional human communication: email, mentoring, leadership advice) can erode trust and credibility in AI.

As Brown put it: “When people use AI for everything, it cheapens the interaction.”

Mandate reskilling, but realise that formal AI education is fragmented

There is no single “AI school,” and the tool landscape is exploding (10,000-plus solutions cited). With that in mind you should focus reskilling on AI literacy and use cases, not tool mastery.

Encourage experimentation across multiple platforms - but within governance boundaries – and expect ongoing learning, not one-time training.

Make IT the ‘Yes’ organisation, with constraints

Every technology problem is solvable if the business accepts the tradeoffs of time, money and people (the “triple constraint”).

First diagnose: Is this a business problem or a technology problem? AI often exposes business process debt, not technical gaps.

IT leadership credibility grows when you frame AI decisions in business terms.

The bottom line for IT leaders

AI is already here. It will change tools, workflows, and org structures, but it will not replace leadership. The leaders who succeed will be those who:

This article originally appeared on our sister site, MES Computing.