AI Called. It Wants Its Quality Department Back.

Why Quality Assurance Professionals May Be the AI Leaders Government Has Been Looking For

Everybody wants AI. The executive says, “We need AI.” The CFO and technology team says, “We bought AI.” The data scientists say, “We built AI.” Then the Quality Assurance professional asks, “Great. But does it work?” And suddenly, the room gets much, much quieter. The objective isn’t to have AI. It is to improve outcomes and drive greater organizational value. This oftentimes includes decisions, cycle time, service delivery, cost, risk, workforce capacity, customer experience or mission-related result sets.

Before asking which AI platform to deploy, leaders should ask a much less glamorous question: What problem are we actually trying to solve?

Here’s the plot twist, AI may be new, but many of its management problems aren’t. AI needs reliable data; QA knows measurement systems. AI needs model validation; QA knows statistical thinking. AI needs process discipline; QA knows process management. AI needs risk controls; QA knows FMEA. AI needs error investigation; QA knows root-cause analysis. AI needs monitoring; QA knows SPC. AI needs governance; QA knows QMS. AI needs improvement; QA knows PDCA and DMAIC. AI needs evidence; QA knows audits and verification.

One of the fastest ways to create an expensive AI failure is to automate a bad process.

Bad process + AI does not equal transformation. It equals bad process at machine speed.

Before automating, leaders should ask what can be eliminated, simplified and standardized; where humans should be augmented; and only then, what should be automated. AI is not a magic box. It operates within a system of inputs, models, outputs, decisions, actions and outcomes, each introducing questions involving data quality, assumptions, bias, human interpretation and unintended consequences.

Government may not need an entirely new management philosophy for every AI initiative. Consider an old friend: DMAIC. Define the problem. Measure the baseline. Analyze what drives errors, bias, variation or failure. Improve through testing, redesign and validation. Control by continuously monitoring performance, drift, risk and outcomes. AI teams may call this lifecycle management. Quality professionals might call it DMAIC with considerably more computing power.

Then there is the new platitude, “keep the human in the loop.” Fine, but which human, and which loop?

What information does that person receive? Do they understand the AI’s limitations? Can they override it? When should they intervene? And who remains accountable? Human oversight has to be designed, not merely drawn onto an architecture diagram. The higher the consequence, uncertainty, human impact or difficulty of reversing an AI-enabled decision, the stronger the case for meaningful human judgment.

The objective should not be maximum automation; it should be the right level of automation for the risk.

AI also changes the trust conversation from “trust me” to “show me.”

Show me how the system was validated. Show me where the data came from. Show me the controls. Show me how performance is monitored. Show me what happens when it fails. Show me who remains accountable.

Operational trust requires evidence, and QA professionals have spent their careers turning “trust me” into “show me.” That creates a larger opportunity for the profession. QA professionals naturally progress in their applicable roles.

Rather than arriving after AI implementation to validate what someone else built, Quality belongs upstream, when problems are defined, requirements established, metrics selected, risks identified and controls designed.

So, at the next AI meeting, ask five WHY questions: What problem are we solving? How will we know AI made the outcome better? What could fail and how would we know? Who remains accountable? How will we continuously monitor and improve it? If the team has thoughtful answers, you may have an AI initiative. If everyone suddenly becomes fascinated with their shoes, you may have an AI Enthusiasm Project.

AI brings speed, scale, prediction and possibility. Quality brings discipline, validation, control, reliability, evidence and assurance.

• Innovation without discipline creates risk.
• Quality without innovation creates stagnation.

Put them together, and you create the conditions for sustainable transformation.

Perhaps the question isn’t whether QA belongs in government’s AI conversation. The question is why we ever thought we could have the conversation without it!

Dr. Rhonda Farrell is a transformation advisor with decades of experience driving impactful change and strategic growth for DoD, IC, Joint, and commercial agencies and organizations. She has a robust background in digital transformation, organizational development, and process improvement, offering a unique perspective that combines technical expertise with a deep understanding of business dynamics. As a strategy and innovation leader, she aligns with CIO, CTO, CDO, CISO, and Chief of Staff initiatives to identify strategic gaps, realign missions, and re-engineer organizations. Based in Baltimore and a proud US Marine Corps veteran, she brings a disciplined, resilient, and mission-focused approach to her work, enabling organizations to pivot and innovate successfully.

Photo Credit: Matheus Bertelli, Pexels