Episode 11: AI Debt Is Going to Be More Expensive Than Technical Debt

If you’ve worked in technology long enough, you’ve probably heard someone say, “We’ll clean it up later.”

Sometimes they’re talking about code.

Sometimes infrastructure.

Sometimes documentation.

Every once in a while, they actually mean it.

Most of the time, “later” never comes.

That’s how technical debt accumulates. One shortcut doesn’t seem like a big deal. Neither does the second. Or the tenth. Eventually those shortcuts become part of how the company operates, and fixing them becomes far more expensive than doing it right in the first place.

I’ve started wondering if we’re about to create something even bigger.

AI debt.

Not because the models are bad.

Not because the technology isn’t ready.

Because AI is learning from organizations exactly as they exist today.

Think about that for a minute.

If your sales process is inconsistent, AI learns inconsistency.

If every region prices deals differently for reasons nobody can explain, AI learns that too.

If customer support has developed five different ways to solve the same problem, depending on who answers the phone, AI quietly absorbs all five.

If your teams spend half their day working around broken processes instead of fixing them, AI doesn’t know those are workarounds.

It assumes that’s how the business operates.

That’s what makes AI different from almost every technology wave before it.

Traditional software automated processes.

AI learns behaviors.

Those aren’t the same thing.

One speeds up work.

The other preserves it.

Including the parts you probably wish would disappear.

I don’t think enough leadership teams are asking a simple question before deploying AI:

“If our AI learned everything about how we work today, would we be proud of what it learned?”

That’s a much more uncomfortable conversation than selecting a model or negotiating a contract.

It forces leaders to confront something they’ve been able to postpone for years.

Some workflows shouldn’t be automated.

Some shouldn’t even survive.

Because every bad habit your organization teaches AI today becomes a little harder to unlearn tomorrow.

Technical debt slows down software.

AI debt risks slowing down the organization itself.

And unlike technical debt, it doesn’t just affect engineers.

It touches every customer interaction, every employee decision, every future recommendation, and eventually, every competitive advantage you hope to build.

The companies that move fastest over the next decade won’t simply deploy AI first.

They’ll take the time to decide what deserves to be taught before the learning begins.