TL;DR: Two loud groups control the AI conversation in our field. One says "developers are finished." The other says "AI is all fake." Both want your clicks, and both stop us from talking about real engineering. Here is what I think we are missing, and what I'm doing about it.
Generative AI has changed a lot since ChatGPT launched in late 2022. In 2025, Claude Code and frontier models like Opus changed how we write code again. Nobody can deny that now.
But something else changed too: how our community talks about its own craft.
We used to be good at this
Developers were early on the internet, and we used it well. We had blogs, books, conferences and meetups. We shared knowledge in public, and we loved to argue about it.
I'm always surprised when I talk to friends in other fields, like lawyers or civil engineers. Many of them don't follow anyone in their profession online. We were different. We had a culture of sharing.
After AI, that culture changed. Not because of the tools, but because of the noise around them.
Two stories, one goal
People who used to talk about architecture, testing strategy and code quality now talk mostly about AI. And they don't talk about how AI changes each step of the software development lifecycle. That would be great. Instead, they share business gossip:
- Which model was just released;
- Which company just (claims to have) replaced its developers with AI;
- Personal stories and reacts of personal stories with former developers who got frustrated with the state of the profession and quit;
- etc;
Two stories came out of this:
- Story #1: the hype. AI is magic, and we will all lose our jobs.
- Story #2: the black pill. Nothing works, every improvement is fake, and the profession is already dead. Some people say the tools don't work at all. Others say nothing good can come from them.
The two stories look like opposites, but they do the same job. They fight for your clicks, and they try to shape how you see your own career.
The result is alienation: we are cut off from the real discussion. To talk about AI, you have to pick one of two sides. That is a false choice.
What we are not discussing
Your team now writes code much faster with AI. Great. But you only win if the rest of the pipeline keeps up:
- Can you do code review for this much code?
- Can you verify that it is ready for production?
- Is your CI/CD strong enough to handle it?
If not, the speed-up just moves the bottleneck from writing code to reviewing it.
But we rarely talk about this. We are busy with "the developer apocalypse" on one side, and "the model is evil, I only use it when my manager forces me" on the other.
Here is what frustrates me most. Everything around the code is code too:
- Infrastructure is code
- Release tooling is code
- Tests and verification automation are code
If AI can generate code this fast, we can use it to strengthen the whole pipeline, not only to ship features faster: better tests, better CI, better observability. The same is true for the recent boom in reverse engineering, where AI helps us understand existing applications, binaries and/or legacy code much faster.
These are the interesting questions. And they matter outside our industry bubble. Society depends on software more every year, so it is in everyone's interest that we build it in a careful, responsible and efficient way.
We can't have that discussion while everyone is stuck in the fight.
Follow the clicks
I suspect many people on both sides don't believe what they say. They say it because, in the attention economy, a click is money.
So social platforms ("the algorithm") stop being a tool and become the goal. Youtube, Instagram and even text social medias like Reddit and Dev were places where people in our field shared ideas and moved the profession forward. Now many people just post what the "algorithm" rewards with views.
It is a reversal. I think we are at its peak, and AI-generated slop makes it worse.
(Yes, I know. I'm posting this on a platform. I used AI to write it. I see the irony. Still, writing it is better than keeping my ideas to myself.)
So what do we do?
For me, it is clear: the internet is no longer the place for these conversations. It can still be a tool, just a different one.
Here is what I'm doing:
- Small private communities with a few engaged people (Discord works fine for that)
- In-person meetups when possible. We lost a lot of this after the pandemic.
- Conversations without pressure from likes, views and retention metrics
That addictive engagement is hurting how we think. I prefer to talk with ten engaged people than to perform for ten thousand people who are just scrolling.
Your turn
Where do you have your best conversations about software engineering today? Is it still online, or did you also move offline?
Tell me in the comments. I really want to know.


