The Real Test of AI in MedTech Is Not the Tool. It Is the Judgment Behind It.

Here is the uncomfortable truth that most conversations about AI in medtech avoid: the organizations that will struggle most with AI are not the ones that move too slowly. They are the ones that move quickly, adopt tools enthusiastically, and discover too late that they never built the foundation underneath them.

That was the thread running through a recent conversation I had with two of the most experienced quality and regulatory leaders in the industry: Kerri DiPietro, Corporate Vice President and Chief Quality Officer at Integra LifeSciences, and Topaz Kirlew, Corporate Vice President and Chief Regulatory Officer at Integra LifeSciences. 

We were joined by more than a hundred medtech professionals for an AdvaMed webinar on AI in quality and regulatory functions. What emerged was a candid, practical, and at times sobering picture of where the industry actually is.

Three things struck me as directly relevant to anyone running a medtech organization right now, regardless of whether quality and regulatory sit in your direct reporting line.

1. AI does not replace judgment. It exposes whether you have any.

The executives leading medtech organizations at this level did not get here by making uninformed decisions. The challenge AI introduces is more subtle than a capability gap. Under deadline pressure, with a polished and plausible-looking output on the screen, the path of least resistance is to defer to it. A human is technically in the loop, but the judgment is not actually being exercised. The point Kerri and Topaz made, repeatedly and clearly, is that we cannot allow AI to substitute for the judgment that got us here.

Topaz described what responsible use actually requires: tracing every material conclusion to an authoritative source, confirming inputs are complete and current, and having a subject matter expert test the output against the real product context, the applicable regulatory requirements, and the known risk profile. AI can accelerate the analysis, she said. It cannot be the final decision maker.

Kerri framed it as a risk-calibration question every leader should be asking: if this AI-assisted decision turns out to be wrong, where and when will we find out? If the answer is usually quickly and internally: in a design review, a DOE result, a validation output, then the decision is low-risk. Higher-stakes decisions would be those that might be discovered external to your organization: in an audit, in a missed safety signal, or in a patient outcome. The level of human oversight applied to any AI output, she said, should be directly proportionate to that answer. That is not a compliance observation. It is a leadership one.

2. The organizations that fall behind will be the ones that adopted tools without building infrastructure.

Kerri was unambiguous on this point. Many companies fail on the AI journey because they only half commit. They want the benefit, but they do not want to create the infrastructure needed to achieve the results. You cannot just adopt a tool, she said. You have to shift how you hire, how you train, and how you govern. All of that has to be built around the capability, not retrofitted onto it.

This is not a quality and regulatory observation. It is a leadership and organizational design observation. The companies that will use AI effectively three years from now are making structural decisions today: about talent profiles, about governance frameworks, about how pilots become standard practice, and about who owns accountability when an AI-assisted decision turns out to be wrong.

Topaz described what responsible governance actually looks like in practice: a monthly AI use case forum within her regulatory team, where people bring ideas, run pilots in a structured way, and build from evidence rather than enthusiasm. The governance enables experimentation, she said. It does not prevent it. When the process is so complex that people work around it, governance has failed.

3. The advice has not changed. The stakes have.

Near the end of our conversation I made an observation that I keep thinking about. Everything Kerri and Topaz recommended for navigating AI, things like humility, mentorship, finding the dissenting voice, challenging outputs, building judgment through deliberate experience, is advice that has always applied to developing strong professionals in regulated environments.

AI has not changed the fundamentals. What it has changed is the consequence of ignoring them.

When the tool generates a polished, plausible, well-structured answer in seconds, the temptation to accept it is real. Especially for professionals who have grown up with these tools and have no reference point for what it feels like to work without them. The judgment required to challenge that output, to know when something that looks right is not right, is not something AI develops. It is something leaders develop, through experience, through mentorship, through deliberate investment in their people.

That is a CEO and CTO conversation. Not just a quality and regulatory one.

Watch the full conversation.

The replay is available here: Quality and Regulatory at the Speed of AI: Moving Faster Without Losing Trust

If this conversation is relevant to something your organization is working through, we would welcome the opportunity to continue it. Reach out to our team at SuntraMedTech.com/contact.