Signal vs Noise In The Time of AI Hype

The Garden

Managing my news feed is like growing a garden. I cannot control every seed that lands in the soil, but over the years I have learned to shape what grows there. Carefully subscribing to channels, commenting on the right topics, occasionally even engaging with content purely to train the algorithm, all of this bends the feed toward something useful. I have been doing this for years and the garden, most days, keeps me informed about the topics I have curated for myself.

But here is the thing about my particular garden: there is no fence between the personal and the professional. When you are a founder of a small consulting company, your work does not end when you close the laptop. The algorithm does not know that at 9 PM I might prefer something lighter, something human, because I have spent years teaching it that I am the kind of person who clicks on AI news at any hour. My feed is a single continuous stream of model announcements, benchmark comparisons, LinkedIn thought leadership and arguments in the comments, and the occasional meme about AI replacing someone’s job. There is no escape hatch. The person and the professional drink from the same firehose of AI news.

This would be fine if the firehose delivered only signal. It does not.

The Paradox

Here is the paradox I live inside: the same flood of AI content that exhausts me is also what keeps me employed.

My clients, managers, directors, business owners across Poland, come to me in part because they have other things to focus on. They are running companies, managing teams, making decisions that have little to do with large language models. But they are also bombarded by McKinsey reports on AI transformation, by conference titles promising the future, by their own social media feeds telling them they are already behind. They need someone to help them separate what matters from what does not. That someone is me, and I am also, on many days, unsure about the answer myself.

I sit in the same deluge. The difference is that this deluge became my profession, which means I cannot simply step out of it. The noise is the price of admission. And some of that noise, processed and filtered, becomes the signal I sell.

This is an uncomfortable position. I am a beneficiary of the AI hype machine. Every breathless announcement, every exaggerated claim about AI replacing entire industries drives demand for what I do. The more uncertain the landscape, the more people need someone to help them navigate it. I am trying to be that person while navigating the same landscape.

Two Wrong Extremes

Two radical responses to this situation present themselves, and both miss the point.

The first: disconnect entirely. Cancel the subscriptions, delete the apps, move to a mountain cabin. The problem is obvious. An AI consultant who does not follow AI developments is not a consultant but a philosopher. My clients are not paying for timeless wisdom. They are paying for someone who knows what happened last week and can tell them whether it matters.

The second: stay plugged in around the clock. Live on Twitter, consume every newsletter, watch every launch event. This is equally destructive, and the reason is economic. The marginal value of information consumption drops off a cliff. The first twenty minutes on the internet each morning might genuinely reshape how I think about a client problem. The next twenty minutes add some texture. After that, the returns collapse. By the one-hour mark, I am not learning. I am scrolling. I am consuming content that exists not because it is true or useful but because someone needed to publish something today.

And yet. Sometimes, deep into a scroll session that should have ended forty minutes ago, I stumble on something genuinely insightful. A thread that reframes a problem I have been thinking about. A demo that changes how I understand a tool. This is what makes it so hard to stop. It is a slot machine. You keep pulling the lever because every so often it pays out, and the occasional reward is just frequent enough to keep you gambling your time away. The house always wins.

John Green captured this dynamic sharply in The Anthropocene Reviewed when writing about 24-hour television news: a format designed to keep you watching inevitably degrades the quality of what it shows you, because the constraint is not “what is worth saying” but “we must fill this hour.” The AI content ecosystem operates on the same logic. The garden grows weeds faster than flowers.

What Moderation Looks Like

So what does moderation actually look like in practice? A set of choices I try to make, with varying success, most weeks.

It starts with continuing to tend the garden, with sharper intention. The internet, for all its noise, remains the only way for me to stay relevant to my clients. And relevance here is not about physical proximity to where AI is being built. I imagine a person living in San Jose, California does not stay magically up to date on AI models just because they share an aqueduct with some model builders. The knowledge travels through the same digital channels I have access to from Poland. The question is how deliberately I use those channels.

It means creating space between consumption and understanding. Riding a bike. Walking. Talking with my wife, a friend, a client about what the things I read actually mean, what they change and what they do not. Both the intake and the digestion matter, but the balance is usually off. Most of us, myself included, consume far more than we process (both in pixels and calories). A thirty-minute conversation about a real AI implementation problem at a Polish firm can teach me more than three hours of scrolling, though I will admit the reverse is occasionally true. It is not always clear in the moment which activity will produce the insight.

It means reading and writing. These are two sides of the same coin. I need good books that explain what is actually happening with AI, the careful kind of thinking that takes months to produce rather than minutes. And I need to write about what I observe, in my own words (often with AI assistance for spelling and a critical eye on the first draft), sometimes spending close to an hour on a single workshop slide. One could argue this is inefficient. I would argue it is the only way to convert noise into signal.

It means showing up in person. If there is a relevant AI event in my part of Europe, I should be there, well prepared and ready to talk with people. The conversations that happen in hallways and over coffee after a presentation have a density of useful information that no feed can match.

And most importantly, it means getting my hands dirty. Actually using the tools I recommend to clients. Solving real problems. Testing different approaches, breaking things, rebuilding them, as my constantly evolving website can attest. The gap between people who talk about AI and people who use AI is wide, and I want to stay firmly on the side of those who use it.

What it does not mean is engaging with the spectacle. The meme wars between AI giants, the gossip drama, the hot takes, the sensationalist coverage. I will not pretend these have zero value. Often a meme captures a truth faster than an essay. But the ratio is not ideal, and treating them as a primary source of understanding is like treating a slot machine as a retirement plan.

Signal Over Noise

Please excuse the performative inconsistency, but I have no reason to believe I will practice all of this perfectly. The deluge is strong. The algorithm is patient and powerful. But there might still be value in naming what good looks like, even when you cannot always reach it.

Signal over noise.

Blazej Kunke, founder of ^Kunke Consulting