AI slop

When I write something, I address it to someone. That person – or people – will spend their precious time reading it. Trying to comprehend. My goal is always to move something further – a process, a task. Unblock something and let someone else get value from what I have to say.

I’ve recently encountered quite a lot of lack of mutual respect. I faced situations when someone (a developer, content creator or a so-far-respected-by-myself writer) throws me a written piece completely generated by AI, hard to comprehend and process. Using words or phrases they never used before. You probably know this feeling very well – someone using basic English for years, suddenly utilizes complex phrasal verbs or weirdly-sounding and out-of-context words. The sentences seem shortened, too complex for what the actual message seems to be.

This is what I consider AI slop. When you work on something, like a research, you get immersed in what you search for. It’s very tempting to use AI to write all of this up. You look at it – looks great, you are in the context so you seem to understand it. Looks clear. Then you throw it towards the team – hey guys, here’s my research outcome. Please read it before our meeting.

There’s usually no reaction at all – no questions asked, no comments on your research PR. Teammates come to the meeting and stay quiet. Someone maybe asks one or two basic questions. Everyone agrees, because your AI-generated fancy HTML research outcome doc looked amazing. They didn’t understand 70% of it, but assumed that there’s probably something wrong with them. That they are too limited to understand it. You are the one who designed this great solution, so everyone just quietly agrees to it. That’s how we end up in AI hell

AI hell

AI slop leads to AI hell. No one knows what’s actually going on in a project. The knowledge is lost somewhere in AI-generated docs (which were never committed, by the way). The problems will emerge much later. Maybe in a year or two. No one will be in the context. There were no discussions when this was designed, no brainstorming, no critical thinking. We’ve lost all of that, because the AI-generated document looked too good.

Do we really want to live in a world like this?

Show some respect

I mean it. When publishing any kind of content – a Slack message, Pull Request description, research outcome document or a JIRA ticket – show respect to your readers. Try to position yourself in their minds. Write that final document by yourself. Or at least review it (at the very least, remove the em-dashes ). Rewrite complicated parts. Use the terms that your team already knows and have been using for years.

If you are short on time, work with your AI agent to make the artifact easy-to-comprehend and record a 5-minutes long screencast where you walk people through your thinking and the proposed solution/PoC. Tell them that the generated document is quite detailed, but here’s a 5-minutes summary so you guys can grasp what I meant with that.

And to be clear – I’m also addressing this article and those remarks to myself. I also found myself guilty of publishing AI-generated content that wasn’t easy to process. But I’m constantly trying to first notice that, and second – improve. We are all in that, but what strikes me is that this shit’s getting real. It’s getting more widespread and the humanity is vanishing from more and more of the content we consume.

Guys, for real. Let’s stop this AI-generated disrespectful bullshit. Let’s use AI for what it’s best for – automating the boring parts and letting us focus on what really matters. Not replacing the human touch in everything we create.

The post Where has mutual respect gone in AI era? appeared first on CodeJourney.net.