Before You Ask What AI Can Do, Ask What Problem It Should Solve

Healthcare has no shortage of problems.

Clinician burnout. Administrative burden. Fragmented information. Delayed access. Workforce shortages. Inefficient processes. Preventable harm.

It also has no shortage of technology promising to help.

That combination makes AI incredibly compelling. A tool arrives with a polished demonstration, an impressive list of capabilities and a vision of work becoming faster, easier and smarter.

The temptation is to ask: What can this technology do?

But that is rarely the best place to begin.

A better question is: What problem are we actually trying to solve?

🧭 Today, We Go Off the Clinical Path…

Technology conversations often begin with features.

Can it summarize? Predict? Recommend? Automate? Generate? Integrate?

Those are useful questions later. At the beginning, they can send an organization searching for places to use a solution rather than deciding whether a problem truly requires one.

When we lead with the tool, almost any inconvenience can become an AI use case. The organization may gain an impressive capability without improving the experience of the clinician, patient or leader who lives with the problem.

Healthcare has seen this before. We implement technology intended to reduce burden, then add new review steps, alerts, documentation expectations and oversight processes. The original task changes, but the burden does not disappear. Sometimes it simply moves to someone else.

A technically successful implementation can still fail to create meaningful value.

🥾 Where the Path Gets Rough

Defining the problem sounds simple. It is not.

Different stakeholders may be experiencing completely different versions of it. Leaders may see delayed performance. Clinicians may see an impossible workflow. Patients may see a confusing experience. IT may see unreliable interfaces. Finance may see avoidable cost.

All of those perspectives can be true.

If the problem is framed too broadly—“documentation takes too long”—the organization may select a tool before understanding which documentation, for whom, under what conditions and why.

If it is framed too narrowly—“we need AI to write this note”—the proposed solution is already embedded in the problem statement.

There is another challenge: some problems are not technology problems.

A broken process does not become healthy because AI is placed on top of it. Unclear ownership, conflicting policies, unnecessary approvals, inadequate staffing and poor communication cannot always be automated away.

AI may accelerate a process. If the process is poorly designed, it may simply help the organization reach the wrong destination faster.

🧭 Check Your Compass

A strong problem statement should be specific enough to investigate and open enough to allow more than one solution.

Try asking: Who experiences the problem? What happens today? How often does it happen? What is the impact? What evidence tells us this is the problem? What have we already tried? What would better look like? What should not be lost while we improve it?

Then ask one of the most important questions in healthcare transformation:

What do the people closest to the work believe is causing it?

The frontline does not have the only perspective, but it has an essential one. People who perform a process every day understand the workarounds, exceptions and invisible steps that rarely appear on a workflow diagram.

Their insight can prevent leaders from solving the visible symptom while leaving the cause untouched.

💡 A Different Direction

Before scheduling the vendor demonstration, schedule the workflow conversation.

Map the current state. Observe the work. Identify where time, attention or information is lost. Separate regulatory requirements from organizational habits. Ask which steps add value and which persist only because they have always been there.

Then define success in human terms.

Will clinicians spend less time searching? Will patients receive faster answers? Will a safety risk be identified earlier? Will leaders have information they can act on? Will the process require fewer handoffs? Will the workforce experience less administrative friction?

Only after that should the organization evaluate whether AI is the right tool.

Sometimes it will be. Sometimes a policy change, integration, clearer role or redesigned process will solve the problem more effectively.

Innovation includes the discipline to choose the simplest solution that creates the needed result.

⭐ Finding True North

AI can help healthcare solve difficult problems. But it cannot compensate for a vague understanding of those problems.

The organizations that gain the most from AI may not be the ones that adopt the greatest number of tools. They may be the ones that ask better questions before choosing them.

What problem are we solving?

Who experiences it?

How will we know life or care is meaningfully better?

And will this technology actually get us there?

Those questions slow the conversation just enough to protect the purpose.

Because the goal is not to bring more AI into healthcare.

The goal is to make healthcare safer, simpler, more connected and more human—and to use AI when it genuinely helps us move in that direction.

Until next time, keep asking questions, challenging what’s always been done and finding your true north. 🧭

Off the Clinical Path—Navigating the challenges, changes and innovations shaping healthcare.