When the Old System Won’t Leave and the New One Won’t Wait

We tend to think about technological change as if it happens in neat stages.

There is the old system. Somebody invents something better. Then, after a little confusion, everybody switches.

That is not usually how it works.

The hardest part of technological change may not be the old system or the new system. It may be the long period when both have to operate together.

Andrew Miller recently described this problem in the context of autonomous vehicles. His argument is that roads may become most difficult not when almost everyone is driving manually, and not when almost everyone is automated, but somewhere in the middle.

Imagine a road filled with conventional human drivers, cars using advanced driver assistance, and fully autonomous vehicles. Human drivers improvise. Automated vehicles follow programmed rules. Semi-automated vehicles do some of both.

The technology may work perfectly well on its own.

The problem is that it has to share the road with everything else.

That may be the real transition problem.

We Already Live This Way

I don’t have to look very far to find examples.

Most of my regular bills are paid online. For everyday purchases, I usually use a credit card because I like having a record of what I spent. Sometimes I use a debit card to avoid the increasingly common three-percent surcharge.

But several people who work around our house still want checks or cash.

And I do not use phone payment systems much because I simply do not like being forced to do everything through my phone.

So in the course of an ordinary month I may use online banking, credit cards, debit cards, checks and cash.

The new payment system did not replace the old one. It added another layer.

Fax may be the strangest survivor of all.

The rotating drums and dedicated fax machines of the fifties are mostly gone, and we have had email, PDFs and electronic document systems for decades. Yet I still get requests from healthcare providers, real estate people and legal offices to send something by fax.

So, I take a digital image of a document, feed it into an app, and the app converts it into a fax for another digital system to receive.

We replaced the technology and kept the process.

Yuck.

Cars May Be Messier

Transportation is going through an even more complicated version of the same thing.

Gasoline cars are going to be around for a long time. I have friends who drive hot rods from the 1930s. They are not waiting for an electric conversion kit.

At the same time, hybrids and electric vehicles are growing. Charging infrastructure continues to expand, but there is still a long way to go before electric vehicles are equally convenient everywhere.

Then we are adding driver-assistance systems and fully autonomous vehicles.

Autonomy will probably spread first in large metropolitan areas. Small towns and rural areas may follow much much later.

Therefore, for quite a while we may have gasoline cars, hybrids, EVs, partially automated cars and fully autonomous vehicles all sharing the same roads.

That is not a clean transition.

It is a pileup of technologies.

Businesses Are Even Worse

The people most enthusiastic about new technology often underestimate how difficult it is to replace something deeply embedded inside a large organization.

The new system looks better. The old system looks obsolete. So why not just switch?

Because the old system is connected to everything.

It has data other systems depend on. It has reports people still use. It has interfaces nobody remembers building. It may contain decades of business rules that were never completely documented.

And perhaps most importantly, it has people who know how it really works.

There are still large companies running mainframe applications whose roots go back decades. Some of those systems may have been modified, wrapped in newer technology and connected to modern applications many times, but the original logic remains buried inside the organization.

Eventually, some of those systems disappear. Sometimes because something better finally replaces them. Sometimes because the last people who really know how to use them retire.

Technology can be replaced faster than organizations can forget how they used to work.

Air traffic control is another example. Everyone has known for years that parts of the system need modernization. But replacing systems responsible for directing thousands of aircraft safely through the sky is not the same thing as installing a new version of Microsoft Office.

The old and new systems often have to operate together for years.

That is when things get complicated.

The New System Often Adds Work Before It Removes It

We usually sell new technology by emphasizing simplification.

Cloud computing will eliminate all those servers. Electric vehicles will eliminate gas stations. Automation will eliminate repetitive work. AI will eliminate tedious processes.

Eventually, some of that may be true.

But during the transition, organizations often have to support both systems. A company moves some applications to the cloud but still maintains older systems. A retailer installs self-checkout but keeps staffed checkout lanes. A business adds AI agents but still needs people to supervise them and people who continue doing the work the old way.

This transition can increase complexity before it reduces it.

That part rarely makes the sales brochure.

AI May Be the Biggest Hybrid System Yet

There is a tendency to assume AI adoption will happen rapidly because the technology itself is advancing rapidly.

I am not convinced.

The Internet seemed to spread overnight when we look back at it now, but it really took more than a decade — arguably two — to become fully embedded in everyday life.

AI asks people to change more than the tools they use. It asks them to change how they work, how they make decisions, what they trust and how much authority they are willing to give a machine.

Some people will adopt AI quickly. Others will use it cautiously. Some will refuse to use it at all.

And those holdouts will not all be seniors like me.

There will be younger people who distrust automated systems, professionals who prefer traditional methods, workers who do not want AI involved in their jobs and consumers who simply do not want another machine making decisions for them.

Thus, the workplace of the future may not be “AI-powered.”

It may be partially AI-powered.

One employee may use AI agents for half of the job. Another may use AI only for research. Another may avoid it entirely.

The company then has to manage all three.

That may create its own version of the autonomous-vehicle problem: different people working according to different assumptions and at different speeds inside the same organization.

Some People May Adopt AI Without Knowing It

There is another possibility.

A lot of people who say they do not want AI may end up using it anyway.

Not because they change their minds, but because AI quietly moves into the products they already use.

It will appear inside cars, search engines, medical systems, customer service, appliances, travel planning, cameras, phones and dozens of other technologies.

The person may never say, “I decided to adopt AI.”

The refrigerator, car, bank or doctor’s office may adopt it first.

Some technologies win because people actively choose them.

Others win because they quietly become part of everything else.

AI may follow something closer to the path of electricity than the Internet: gradual infrastructure buildout, uneven adoption and long periods where old and new systems coexist.

Maybe the Transition Never Ends

Andrew Miller’s argument about autonomous vehicles contains one more uncomfortable idea: the mixed period may not be temporary.

Human-driven vehicles may never disappear completely. Emergency vehicles, construction equipment, rural drivers, hobbyists and others may continue to operate alongside autonomous systems indefinitely. If that is true, then the goal should not necessarily be to design for the day when everything is automated. The goal should be to design systems that work when different technologies coexist.

That may apply to much more than transportation. Cash may never disappear completely. Gasoline vehicles may remain for decades. Human customer service may exist alongside AI. Old corporate software may survive inside organizations long after everyone agrees it should be gone, and AI may become deeply embedded in business without ever becoming universal.

The future may not be one new system replacing the old one.

It may be permanent hybridity.

The Messy Middle May Be the Future

We tend to imagine technological progress as a destination.

Eventually the new thing wins. Eventually everybody adopts it. Eventually the old system goes away.

History suggests otherwise.

The old system usually hangs around longer than expected. The new system arrives faster than institutions can absorb it. And the rest of us spend years — sometimes decades — figuring out how to make them work together.

That may be where the real challenge of technological change lives.

Not in inventing the future.

Not even in adopting it.

But in surviving the long period when the old system won’t leave and the new one won’t wait.