The Atlassian Friend #9: From Pair Programming to Pair Working

[#1]

I was afraid of AI.

I think that’s probably the right place to start this article.

Not curious-but-cautious. Not professionally skeptical.

Afraid.

I had spent almost thirty years working in technology, learning, adapting, solving problems, building things, breaking things, fixing them, and starting again.

And suddenly, here was something that could write code, explain concepts, analyze problems, generate documentation, propose architectures, create diagrams, and do in seconds things that had taken me hours earlier in my career.

It was difficult not to wonder:

What happens to people like me?

What happens to all those years spent learning how to do this?

Could AI eventually make all of that obsolete?

I don’t think those were unreasonable questions.

What I didn’t expect was that actually working with AI would gradually change the questions themselves.

• • •

When I started in technology, learning looked very different.

You read.

You learned.

You tried.

It didn’t work.

You tried something else.

Sometimes you had documentation.

Sometimes you had a book.

Sometimes you had an example that was vaguely related to what you were trying to accomplish.

And sometimes you had almost nothing.

I remember the feeling more than any particular technology.

You knew what you wanted to achieve, but you had never done it before.

It could feel a little like trying to invent Coca-Cola without knowing the formula.

You had some ingredients.

Maybe some clues.

Perhaps an example of one small piece.

Then came experimentation.

A lot of it.

Trial and error wasn’t some formal learning technique.

It was simply how we worked.

And eventually, after enough reading, testing, failing, understanding and trying again, something worked.

That feeling was wonderful.

Then the Internet arrived.

And that really was a revolution.

Suddenly, knowledge wasn’t limited to the books on your desk or the people around you.

Search engines.

Forums.

Mailing lists.

Technical communities.

Later, countless websites, blogs and places where developers and administrators shared what they had learned.

You could discover concepts you didn’t even know existed.

Someone would mention something unfamiliar, you would search for it, find examples, read about it, discover another concept along the way and keep going.

Learning became dramatically faster.

But there was still a familiar cycle.

Search.

Read.

Find something that looked promising.

Try it.

Maybe it worked.

Maybe it almost worked.

Maybe it failed completely but gave you enough information to understand what you needed to search for next.

Then you went back.

Search again.

Read again.

Try again.

The Internet gave us incredible access to other people’s knowledge.

But we were still the ones doing all the synthesis.

We had to decide which answer applied.

Which one was outdated.

Which example was close enough.

Which concept mattered.

Which pieces could be combined into something that solved our particular problem.

For many years, that became normal too.

And then AI arrived.

• • •

At first, I think I approached it much like another way of finding answers.

Ask something.

Get something back.

Useful.

Interesting.

Sometimes impressive.

But gradually, without any particular moment when I consciously decided to change how I worked, something else started happening.

I stopped only asking questions.

I started having conversations.

I would explain what I was trying to accomplish.

I would propose an idea.

AI would propose another one.

We would compare them.

I would explain why one of the options wouldn’t work in my environment.

We would change it.

Then I would try something.

It would fail.

I’d bring back the error.

We would debug it.

Try again.

Sometimes the proposed solution would be wrong.

Sometimes my idea would be wrong.

Sometimes we would go several rounds before realizing that the approach itself needed to change.

And that felt strangely familiar.

I had experienced something similar before.

Pair programming.

Pair programming was never really about two people taking turns typing code.

The interesting part was having someone beside you while solving the problem.

One person sees something the other doesn’t.

Someone proposes a direction.

The other challenges it.

You talk.

You test.

You fail.

You reconsider.

You keep going.

Except today, programming is only one small part of what I do.

I may be thinking about architecture.

A Jira Service Management design.

Forge.

An API.

Automation.

A business process.

Data.

A diagram.

An article.

A problem I can’t quite explain yet but know doesn’t feel right.

And increasingly, I approach all of those things in much the same way.

I explain what I think.

Something comes back.

I react to it.

We change it.

Sometimes I build a first version.

Sometimes it breaks.

Sometimes I come back with an error and we discover the problem isn’t where either of us expected it to be.

And sometimes, after several attempts, the best decision is simply:

No.

We’re going back to the last version that worked.

That doesn’t feel like asking a tool for an answer anymore.

It feels closer to working through the problem with something beside me.

So perhaps pair programming isn’t quite the right description anymore.

I’ve started thinking of it as something closer to:

pair working.

• • •

I want to be careful here.

AI does not work for me.

And it certainly doesn’t do my job for me.

It complements me.

That’s become an important distinction in the way I think about it.

I bring the context.

The experience.

The understanding of the environment.

The history.

The constraints.

The instinct that tells me that although something looks technically correct, it doesn’t feel like the right architecture.

AI can help me search.

It can process an enormous amount of information.

It can connect things.

It can propose possibilities.

It can give me something to react to.

And that last part is perhaps more important than it sounds.

With search, I got things to read.

With AI, I can get something to challenge.

“No, that won’t work because…”

“What happens if we do it this way instead?”

“Compare these two options.”

“There’s something wrong with this design. What are we missing?”

“Let’s try it.”

“It failed. Here’s the error.”

That changes where the work begins.

AI didn’t remove my work.

It moved the starting point.

And it accelerated almost everything around it.

Learning something unfamiliar.

Exploring an idea.

Comparing approaches.

Building a first version.

Debugging it.

Explaining it.

Turning something in my head into a diagram.

Taking data and seeing it visually.

Testing whether an argument makes sense before presenting it to someone else.

The distance between having an idea and being able to examine that idea has become much shorter.

And that means something more important than simply doing more work.

My time can be spent differently.

Less time searching for the ingredients.

More time deciding what to cook with them.

Less time producing the first representation of an idea.

More time questioning whether the idea is any good.

Less time trying to remember how something is done.

More time asking whether it should be done that way at all.

• • •

There is another part of this that matters to me.

When you spend many years working in technology, what you accumulate isn’t simply technical knowledge.

Technology keeps moving.

You learn one language.

Then another.

New frameworks appear.

New platforms.

New methodologies.

New ways of working.

New tools.

And every once in a while, you find something that changes more than the way you perform a particular task.

For me, Atlassian was one of those things.

I started my career as a developer.

I had no idea that Jira, Confluence and eventually the Atlassian ecosystem would become such an important part of where my career took me.

It wasn’t a plan.

It happened along the way.

You discover something.

You learn it.

You use it.

You solve a problem with it.

Then another.

You become curious.

You go deeper.

And sometimes, without really noticing when it happened, something that started as another tool becomes part of your professional journey.

But technology isn’t only languages, platforms or tools.

It is also everything around them.

The methodologies you learn.

The projects you work on.

The mistakes you make.

The roles you take on.

The communities you become part of.

And the people you meet.

People who teach you things.

People who challenge you.

People who encourage you.

People who trust you with something you haven’t done before.

People who introduce you to an idea or an opportunity that changes where you go next.

That is technology to me.

Not simply the technology itself.

The technology and its environment.

And over time, all of that becomes experience.

Experience isn’t a list of technologies you once learned.

It is everything you carry with you after years of learning, changing, making mistakes, meeting people, trying different roles and seeing problems from different angles.

That was something I didn’t fully understand when I first started worrying about AI.

I was thinking about experience almost as if it were a collection of knowledge.

Something AI might eventually know too.

But experience isn’t only what you know.

It is also how you arrived there.

What you have seen.

What you have tried.

What failed.

What worked.

Who helped you.

What changed your mind.

And the perspective you picked up along the way.

And then AI arrived.

At first I saw it as something that might invalidate part of that journey.

Now I increasingly see it as the next thing I have to learn how to work with.

• • •

One of my fears was that AI would reduce the value of what I knew.

My experience told me almost the opposite.

The faster an answer arrives, the more important it becomes to know whether it is a good answer.

AI can propose something very confidently that is completely wrong.

It can misunderstand an important constraint.

It can take us down a path that looks elegant and then collapses the moment it meets reality.

I’ve seen that happen.

Quite a few times.

And the answer isn’t to stop using it.

The answer is to work with it.

Challenge it.

Test it.

Verify it.

Know when something doesn’t feel right.

And sometimes say:

“No. We’re going back to the last version that worked.”

That is why I don’t believe AI has made experience less valuable.

If anything, I find myself using more of my experience because I can move through possibilities much faster.

Experience gives me perspective.

I have seen technologies arrive that people thought would change everything.

Some did.

Some didn’t.

I have seen tools become essential and later disappear.

I have learned things that became obsolete.

And I have learned things that stayed useful even when the technology around them changed completely.

AI is obviously different in important ways.

But the need to adapt isn’t new.

Technology has always asked that from us.

AI made answers cheaper.

It didn’t make judgment cheaper.

I still own the decision.

And I think I always should.

• • •

But perhaps the most unexpected change hasn’t been productivity.

It has been curiosity.

After enough years in technology, you never stop learning, but learning can become very practical.

You learn the new thing because you need it.

You read the documentation because you have a problem.

You investigate the technology because a project requires it.

AI has made the cost of following a curiosity incredibly small.

Something appears during a conversation that I don’t know.

“What is that?”

That answer introduces another idea.

“Wait. How would that work here?”

Then another.

“What if we combined these?”

And suddenly I am exploring something I hadn’t planned to learn at all.

Sometimes it is something I probably would have ignored a few years ago.

Not because it wasn’t interesting.

Simply because finding out whether it was interesting would have required an afternoon.

Now I can look through the door.

Sometimes there is nothing useful on the other side.

Fine.

But sometimes there is.

And then I keep going.

The easier an idea becomes to explore, the more ideas you’re willing to explore.

And I’ve noticed something in myself that I didn’t expect.

AI has made me curious again.

Not because I had stopped learning.

But because exploring feels playful again.

And maybe there is something familiar about that too.

Because technology has been doing this to me for almost thirty years.

A new language.

A new platform.

A new tool.

A new person who shows me something I hadn’t seen before.

A new way of thinking about the same problem.

For a technology I was once afraid might make me obsolete, that is a strange outcome.

It is making me want to learn more.

• • •

I still have questions about AI.

Lots of them.

I don’t know exactly what it will do to our profession.

I don’t know what working in technology will look like ten years from now.

And I certainly don’t think every concern I once had has magically disappeared.

But my relationship with AI has changed.

I no longer see it primarily as something that might replace the things I learned to do.

I see it increasingly as something that allows me to do more with what I have learned.

And perhaps that’s the part I got wrong at the beginning.

I thought AI might diminish the value of nearly thirty years of experience.

Instead, it seems to be giving that experience leverage.

Because those thirty years were never just thirty years of technical knowledge.

They were thirty years of change.

Learning.

Adapting.

Finding new tools.

Meeting people.

Getting things wrong.

Changing roles.

Discovering things I didn’t know I needed to know.

Slowly becoming the person who approaches problems the way I do today.

AI doesn’t replace that journey.

It becomes part of it.

I sometimes wonder what the younger version of me would think if he could sit beside me for a day and watch how I work now.

The books are mostly gone from the desk.

I don’t spend hours searching through pages hoping that one of them contains the missing piece.

I don’t build every first version manually.

I can have an idea, talk about it, challenge it, visualize it, build something, break it and debug it — sometimes in the time it once took me just to understand where to begin.

But I suspect he would also recognize quite a lot.

I still try things.

I still get them wrong.

I still break things.

I still go back.

I still learn.

Underneath all of it, I’m still doing the same thing I was doing all those years ago.

Trying to understand a problem well enough to solve it.

The languages changed.

The tools changed.

The way we worked changed.

My role changed.

And somehow, I kept changing with them.

Maybe AI is another one of those changes.

A very big one.

The difference is that these days, I don’t always have to think through it alone.

We used to call something similar pair programming.

Maybe now I’m learning pair working.

[Debbie Lindsey]

What a beautiful journey!