The doom prophets in your feed are human model collapse

RIP PowerPoint. RIP McKinsey. RIP Your Job.

This is the battle cry from the "Death-by-AI Prophets" I'm seeing everyday in my LinkedIn feed.

It's a form of an intellectual doom loop. An echo chamber phenomenon.

The playbook is always the same. A flood of posts, all born from the same "viral post prompt" based on the same set of marketing non-sense messages fueled by billions of VC & PE money. Not original. Not insightful. Just… boring.

But here is what the prophets don't understand. They are a symptom of a process researchers call "Model Collapse."¹

A study by researchers at Oxford and Cambridge found that when AI models are repeatedly trained on AI-generated content, they begin to forget reality. Their output loses diversity, becomes dumber, and eventually hallucinates a distorted version of itself.

We are witnessing the human equivalent of this. Live in our feeds. A conversation collapsing into a single, repeating echo. A tidal wave of content optimized for the algorithm, not for insight.

This isn't just bad for our LinkedIn feeds. It's a strategic threat to business.

Because in a world flooded with AI-generated mediocrity, the rules have changed.

Where mediocrity is abundant, your judgment is the only asset. Where hype is deafening, your discernment is the only compass. Where everyone is regurgitating slogans from OpenAI, Microsoft, and Google, your original thoughts are the only true signal.

To ignore this is not a strategic error. It is a failure of leadership that will lead to a failure of humanity.

Our job is to be the architects of our future. Not the victims of a narrative written by someone else.

¹ Shumailov, I., Shumaylov, Z., et al. (2023). The Curse of Recursion: Training on Generated Data Makes Models Forget. University of Oxford, University of Cambridge.

Sources

  1. 1: Shumailov, I., Shumaylov, Z., et al. (2023). The Curse of Recursion: Training on Generated Data Makes Models Forget. University of Oxford, University of Cambridge.