The Age No One Knows Again
Chapter 12
Change usually came in numbers before announcements.
There was no email saying we would reduce staff, and there was no notice about reorganizing the development team.
Instead, small decisions followed one after another.
Hiring was postponed.
One person we were supposed to hire disappeared.
One external contractor's agreement ended, and there was no renewal.
The remaining people took on the work that was missing from other teams.
Someone resigned, but their position was not immediately filled.
The team lead said similar things in every meeting.
"For now, it seems we'll have to work with the current staff."
At first, I didn't think much of it.
That's just how companies were.
Sometimes we hired a lot of people, and sometimes we cut back.
When the economy was bad, we reduced hiring, and when projects ended, external staff left.
I had seen this many times.
I thought this was just another case of that.
Then one day, the team lead explained the sprint plan.
The next month's task list was displayed on the screen.
The work to be done hadn't decreased.
In fact, it had increased.
Someone asked.
"Are we doing all of this this month?"
The team lead nodded.
"It looks like we need to do everything with high priority."
"What about staffing?"
"Adding more staff is difficult for now."
The conference room fell silent.
The team lead continued.
"Instead, we've been using a lot of AI tools lately, and actual productivity has been going up, so we've factored that in when planning."
Someone said with a laugh.
"Does AI do the work of one person?"
The team lead laughed too.
"I'm not sure about a full person."
It passed off as a light joke.
I laughed too.
But I looked at the task list on the screen again.
In the old days, we would definitely have assigned more people to this amount of work.
The junior developer next to me said in a low voice.
"I wonder if we can actually get all of this done."
"We'll just have to try first."
I answered out of habit.
After the meeting ended, I returned to my seat.
A new ticket had been assigned.
It was a data cleanup task that had been postponed for a long time.
There were many target tables, and old logic was mixed in.
The estimated effort was four days.
I laughed for a moment.
A month ago, four days would have been correct.
Now it might be finished in a day and a half.
I just had to show the AI the table structure, have it explain the relevant queries, and ask it to create the conversion logic.
I started the work.
It was actually fast.
Analysis was complete by the morning.
Code came out in the afternoon.
Testing was finished the next morning.
It was more than a day faster than expected.
The team lead sent a message.
Are you done with this already?
Yes. We're testing it and it looks like there are no major issues.
A reply came soon.
As expected, there's a big difference when using AI.
I replied Yes haha.
It was praise.
There was no reason to feel bad.
That afternoon, there was a full development meeting.
It was a place to share materials from the upper organization.
Numbers appeared on the projector screen.
Development productivity.
Automation rate.
Deployment frequency.
AI tool usage rate.
In the past, we would have discussed code quality or the number of incidents more, but these days productivity was mentioned frequently.
One person can do more work.
We create more features with the same number of people.
Repetitive work is automated.
The materials contained many such sentences.
The presenter said.
"Teams with high AI utilization are definitely seeing increased throughput compared to the same number of people."
Someone asked.
"Then is that reflected in staffing plans?"
The presenter paused briefly.
"Of course, we need to consider that in the long term."
The conference room became a bit quiet.
The presenter immediately continued.
"That doesn't simply mean reducing people. The direction is that existing staff can focus on higher-value work."
I knew that sentence.
It was something I heard every time automation was discussed.
Machines do repetitive work,
and people do more important work.
It was a nice sentiment.
The problem was whether there was enough important work for everyone.
After the meeting, I met a developer from another team in the hallway.
He was an external developer a few years younger than me.
He had been working at the same company for several years.
"Hey."
"What?"
"My contract ends this month."
I didn't understand for a moment.
"What do you mean?"
"My contract."
"Why?"
He shrugged.
"They're cutting the budget."
"Have you found another place to go?"
"Not yet."
He laughed.
"Well, I'll just have to look for another one."
His words were light, but his face was not.
"You've been here for a long time."
"That's why I thought I'd be here longer."
He laughed.
I didn't know what to say.
The words "it'll be fine" didn't come easily.
At our age, finding a job wasn't as simple as it used to be.
Fewer places hired based on a single skill, and there was also the issue of salary and age.
"Let me know. If I hear of anything, I'll look into it for you."
"Thanks."
He left first.
I returned to my seat.
The sentence I saw in the meeting earlier came to mind.
Existing staff focus on higher-value work.
What kind of value did he have that was considered low?
He knew the old systems well and was always called first when problems occurred.
He did similar work to me.
That made it even more uncomfortable.
A few days later, the junior developer came to my seat.
"Senior, can you take a look at this for a moment?"
There was code created by AI on the screen.
"What is it?"
"It's data transformation. It runs, but the results are a bit odd."
I looked at the code.
One condition was missing.
"This is missing an exception condition."
"Where?"
I pointed with my finger.
"This customer group needs to be handled separately."
"Why?"
"Because of an old policy."
"Is it in the documentation?"
"Probably not."
The junior developer laughed.
"Is it another one of those?"
"Yeah."
I explained.
Why that policy was created.
When it started being applied.
What else needs to be checked if anything is changed.
The junior developer took notes.
"I wish we could put things like this in the AI too."
"I guess so."
"Then next time we wouldn't have to ask."
The junior developer said this without much thought.
I nodded too.
"I suppose so."
After that friend went back to their seat, I stared at the screen for a while.
It was the right thing to say.
It would be nice to document what's only in my head and make it searchable by AI.
It was also good for the company.
Systems that depend on specific people were dangerous.
Even when someone leaves, the knowledge should remain.
I had said that countless times too.
We need to document it.
Knowledge should stay in systems, not in people.
That was right.
But if I pushed that argument all the way, one strange question remained.
Even after everything I know goes into the system,
will the company need me as much as it does now?
I shook my head.
I told myself it was a useless thought.
There was still much work I had to do.
Even if AI gives good answers, people still have to judge which answer is correct.
Old systems had more exceptions than documentation.
Reality doesn't organize itself that easily.
At least not now.
At the team meeting on Friday that week, the plan for next month came out.
The team lead spoke while clicking through the slides.
"Headcount will remain the same. We have no plans for new hires."
Someone asked.
"No external hires either?"
"Right. That seems difficult for now."
The team lead paused and added.
"Instead, please automate everything you can, and actively use AI."
I looked at the screen.
A few months ago, I would have been happy to hear that.
Because it meant I could use new tools freely.
I didn't dislike it now either.
AI was still fun and actually helpful.
The problem was that I was using it too well.
Work sped up.
The amount one person could handle increased.
The company knew that too.
And of course, the company calculated.
Whether they could do the same work with fewer people.
I couldn't even say that calculation was wrong.
I would have thought the same.
An AI window was open on one side of my laptop screen.
I pasted the code from my last task of the day.
Find the unnecessarily repeated parts in this code.
An answer came back in a few seconds.
AI precisely identified the repeated parts.
I deleted them one by one.
The code became shorter.
It became simpler.
For a moment, I thought: weren't people's jobs being organized the same way?
Finding duplicates,
Cutting out the unnecessary,
Creating the same results with fewer resources.
I stopped that thought.
I thought code and people were different.
Of course they had to be different.
As I left work, I looked around the office.
I saw a few empty desks.
Some desks, I couldn't immediately remember who used to sit there.
When someone left, the empty desk was noticeable for the first few days.
After a while, you got used to it.
Becoming a developer with good tools,
and remaining a necessary developer
were not the same thing.
