LinkedIn Is Downranking AI Slop — And I'm Taking a Personal AI Detox

Back in February, I warned that AI-generated phrasing is increasingly perceived as “AI slop” by readers, and that it makes absolute sense to consciously avoid AI in certain contexts, like LinkedIn posts, communication with colleagues, or high-stakes business writing.

Unfortunately, the same now applies to AI-generated images, as magical as the new capabilities of tools like GPT-Image-2 may feel.

On X/Twitter, this has already gone so far that AI expert voices are now deliberately using poor spelling, clunky grammar, and a messy layout to distinguish themselves from the sea of polished generic content. It is a fascinating and somewhat absurd social phenomenon: we have reached a point where grammatically perfect writing is subconscious shorthand for “bot,” while typos and “GenZ speak” have become the ultimate proof of human origin.

Our human pattern recognition is incredibly sharp. We have quickly developed a collective, subconscious radar for the “perfectly written fake.” Paradoxically, the very things we used to get points deducted for in school—informal phrasing, fragmented sentences, and raw, unpolished writing—now feel like the only things that are authentic and human. On the flip side, traditional stylistic devices like em-dashes or structured ellipsis rules are suddenly flagged by readers as AI-typical.

It seems LinkedIn has now caught on to this shift.

Karissa Bell, writing for Engadget, reports that LinkedIn is actively changing its algorithm:

• Reducing the reach of “AI slop”: Generic AI posts with no added value are being targeted.
• Focusing on engagement bait: The platform is cracking down on fake thought leadership.
• Detecting phrasing patterns: Typical AI-typical conversational style and standard phrases are being recognized.
• Prioritizing originality: Real expertise and unique angles are being pushed.
• Keeping helper tools allowed: AI-assisted content is still permitted, provided it contains a personal perspective and real value.

While this sounds promising, some healthy skepticism is warranted. LinkedIn has made algorithmic promises before, and the reality in our feeds didn’t always match the press releases. How exactly do they plan to distinguish between AI-assisted content with genuine substance and pure, uninspired “slop”?

There is a real danger of an algorithmic arms race here. If the platform only filters for semantic patterns, creators will simply learn to adapt their prompts to generate “human-looking slop.” We end up spinning in circles of semantic camouflage where the style changes, but the lack of substance remains. Furthermore, LinkedIn itself heavily promotes its own AI writing assistant, creating an inherent contradiction: the platform is penalizing the exact standardized tone its own built-in tools help generate.

There is also a valid concern about linguistic diversity. If benign writing habits or post-editing refinements get caught in the algorithmic crossfire, we might end up restricting how we express ourselves just to appease a machine.

My Personal AI Detox

The real issue, however, goes deeper than algorithm updates.

Through my own regular AI use, I noticed a creeping laziness. I became lazy in formulating texts properly, lazy in keeping up with my spelling, and—most dangerously—lazy in thinking through complex concepts in depth. These happen to be the exact cognitive strengths I am employed for. (The laziness was of course always there, there just used to be no alternative but to pull yourself together)

The thought constantly hovering in the back of my mind was: “Ah, the AI will polish this draft anyway, no need to sweat the details.” But when you outsource the formulation, you slowly begin to outsource the thinking. The “thinking muscle” atrophies faster than we care to admit. It’s similar to how we stopped memorizing phone numbers when smartphones arrived, or how our sense of direction faded with GPS.

I realized I was falling into a validation loop. Part of the “AI addiction” is that using these tools makes us feel incredibly capable and productive in the moment, regardless of whether the actual output has any real depth.

So, I pulled the emergency brake. I uninstalled my convenience AI helpers like Raycast and am now working (temporarily) completely without AI. Or at least, I am trying to.

To keep my brain from taking the easy way out, I am focusing on what I call “analog-first” workflows. Many of us are finding that writing thoughts down with pen and paper before ever touching a screen is the only way to slow the brain down, connect with the material, and ensure that our own perspective remains the driving force.

The only compromise I still allow myself is recording raw voice memos, which I then have AI clean up and structure into readable text blocks.

Rediscovering the Balance

This detox isn’t about being a Luddite. AI is a tool, and when used correctly, it is incredibly powerful. The real distinction lies in whether we use AI to replace our thinking or to amplify it.

The problem is the aimlessness of the user. AI didn’t create shallow content, it merely scaled what was already there: massive volume with zero substance. AI slop is essentially the use of technology to simulate depth rather than to express genuine expertise faster.

When you use AI to condense your own unique thoughts, raw notes, or voice memos into their essence, the result is highly valuable. But when you use a five-word prompt to inflate a thin idea into a 500-word essay, you are creating the exact noise that makes people want to log off.

A fascinating example of this shift comes from the software development world: some open-source maintainers have started asking contributors to submit the prompts they used rather than the finished, AI-generated code. Why? Because the prompt is where the actual human intent, logic, and core concept lie. The code is just the execution and the maintainers have their own proven workflows for that.

Historically, this debate isn’t entirely new. When the Gutenberg press was invented, critics feared that standardization would kill the unique art of the scribes and ruin the transmission of knowledge. Instead, knowledge became scalable. We are witnessing a massive shift in how genuine expertise is defined and perceived.

At the end of the day, the dose makes the poison. Ultimately, those who choose to think more in the age of AI—rather than less—are the ones who will stand out.

What is your current zeitgeist on this? Are you scaling your thinking, or are you ready for a detox of your own?