Learn Why “Inpatient” and “Observation” are Insufficient Indicators

And I don't know if I have anything to say about the idea of AI taking over the world and killing all of us.

But I did want to talk about an incident that happened and the growing AI dependence on the skills that we risk losing.

So when part of my chat GPT account stopped working, I signed up for Claude to help with an Excel task.

My reaction raised a question about how much we are learning to depend on AI.

So after more than 40 years in healthcare and accounting, I found that reaction was worth examining.

I had reached for another AI instead of doing it myself.

And it reminds me of the map that we no longer unfold.

So think about GPS.

So how many of us could take a paper roadmap, locate ourselves, plan an unfamiliar route without reaching for the phone? We can follow turn-by-turn instructions and arrive successfully without ever really learning the roads, how the roads connect.

When the directions disappear, we discover how little of the route we actually remember.

So AI can supply the reasoning too.

So AI can go further than calculating a number or suggesting a route.

We can ask it to choose an approach or construct a formula or explain a result.

If we accept that explanation without understanding it, we may finish the assignment without being able to support it when asked.

And I've already had the experience of asking someone about an Excel formula only to realize that they have no idea how it was created.

Research offers reasons to pay attention.

A 2025 survey from Microsoft of 319 knowledge workers found that the greater confidence and found a greater confidence in generative AI was associated with less critical thinking.

The researchers also found that AI shifted critical thinking towards checking information and integrating results.

There were self-reporting findings from Microsoft team and not really proof that AI permanently reduced a person's ability to think.

But it does make you think.

So why does this matter in healthcare finance? In reimbursement and compliance work, completing a spreadsheet is only part of the assignment.

Someone may understand which records belong in the calculation, what the formula measures, and whether the result is supported by the applicable rules.

So consider someone using AI to repair a reimbursement workbook or a list of claims or something that they're looking at as claims analysis.

A formula might calculate successfully while including the wrong population, excluding relevant records, and the analyst needs to understand and have enough knowledge of the data to recognize errors.

I'm especially concerned about newer staff on that judgment.