AI writes. Bots engage. Algorithms rank. Honestly — when did we lose sight of human beings as the end consumers?

For a few weeks now, I've been dwelling on a question that won't let me go. Not because it's new — but because I think that as an industry, we're collectively pretending it isn't.

“The question is simple: Who are we actually doing this for right now?”

— Olivia Ulbing-Sommeregger, Founder oha! digital

I see it and do it myself every day. Content optimised for AI search engines. Engagement generated by Manychat. Social media strategies built on algorithmic logic,  not on human curiosity. And somewhere in this loop, something very fundamental has shifted — without us really voicing it out loud.

We are increasingly communicating from machine to machine. And calling it innovation in marketing.

The bot-to-bot loop — what's actually happening right now

Let’s get into details. Facts and figures, because I'm no fan of abstract warnings.

According to a 2024 analysis by Imperva, 49.6% of all internet traffic now comes from bots — for the first time in history, more bots visit our websites than humans. A growing share of this consists of so-called “bad bots”: automated accounts that like, comment, and share — thus inflating engagement metrics that we then sell as success.

At the same time, content production has exploded thanks to generative AI. A 2024 study by NewsGuard identified over 1,000 (Cute, right? Should I ask Claude how many there are in 2026?) websites publishing entirely AI-generated content — without human editorial oversight, without context, without added value. Just output. Just volume.

And then there's the side that has everyone occupied right now: AI search. Google SGE, Perplexity, ChatGPT Search. Nobody is optimising for the person googling anymore — but rather for the AI that summarises the answer. Which means: we write content so a machine can read it, so it can provide a summary to a human.

The human comes third in this chain.
Crazy, right? And no, this isn't meant to be AI shaming. I wouldn't know where to find all the extra hours in a month to handle all the agency tasks at the same pace. What I can't get out of my head right now is the fact that, in my opinion, we are making a huge category error: we are confusing optimisation with communication.

The engagement problem nobody dares to talk about

When I look at current social media reporting — from clients, from competitors, across the industry — I see engagement metrics that look great at first glance. Likes. Comments. Saves. Shares.

And then I take a closer look. Well you know me – right?!

A growing proportion of this engagement comes from accounts that are either automated or powered by AI tools designed to simulate “authentic engagement” (fake it 'till you make it, Claudy). Entire categories of tools have emerged that promise exactly that. Your content gets interactions — from whoever, as long as the number looks right.

The problem: algorithms reward this in the short term. Reach increases. The report looks great. And nobody asks whether there was an actual human being anywhere who looked at it and thought: “This is relevant to me.”

According to a 2023 study by HypeAuditor, an average of 55% of Instagram influencer accounts with over 100,000 followers exhibit suspicious follower growth — an indicator of bots or purchased engagement. This isn't a niche phenomenon. It has become an industry standard. Sadly.

And what about us as agencies? We celebrate this in our reports. We don't question why that one post with the trendy hashtag had so many views. We build strategies based on numbers that, in part, don't even come from humans.

AI content: The promise of volume and its costs

I get the appeal. Truly.
Generative AI can produce a hundred blog articles in the time it takes me to rewrite this piece. Ten landing pages. Thirty social media posts. That's not an exaggeration; that's the technical reality of 2026.

And for many businesses, it sounds like an all-in-one silver bullet. It solves one or more problems simultaneously: too little content, too little time, too little budget.

The issue isn't the volume. The issue is what gets lost in the process.

Character.

AI has no opinions. It has patterns. It can simulate what sounds like an opinion — but it doesn't know why you, as a company, do what you do. It doesn't know the moment your team decided to give a client honest advice instead of telling them what they wanted to hear. It can't write what you haven't inputted.
Specificity. An AI-generated article on “Social Media Strategy for SMEs” sounds like every other article on “Social Media Strategy for SMEs”. Because it was distilled from the exact same sources as all the rest. The differentiating advantage — your lived experience, your concrete case studies, your perspective — falls by the wayside.

Trust.

And that is the point that concerns me the most. A 2024 study by Edelman shows: 71% of B2B buyers say that thought leadership strengthens their trust in an organisation — but only if it's authentic and substantial. Generic AI content does the opposite. It signals: We haven't really given this thought. We just produced. Pure actionism.

The algorithm is not your customer

What we tend to overlook amid all the hype around vibe coding, content production, and the agentic web: The algorithm doesn't buy anything. At least not yet.

It doesn't book services.
It doesn't pass on a recommendation.
It doesn't sit in a meeting saying, “I think we should work with this agency.”

Humans do that.
And humans operate on a different logic than algorithms. They're not looking for the most optimised text. They're looking for the feeling that someone genuinely understood their problem. They're looking for humans, for empathy — ultimately, all of us just want to be understood.

AI content can deliver that if it's well-crafted and curated with a human touch. But it cannot deliver it when produced on autopilot and distributed via bot engagement.

I work every day with brands making this exact mistake right now. Not out of malice — but simply because it's genuinely cool what we can do ourselves these days. I get that. But efficiency without impact is not a win. It's merely being busy without moving forward.

AI content marketing for the win?

I'm not anti-AI.
We use generative AI at oha! digital — for research, for initial drafts, for background automation. AI is a tool, and I believe it's wrong to moralise tools.

However, I believe it's right to state clearly where I draw the line:

1. Content written for humans must actually interest real humans. Not look like an SEO checklist disguised as continuous text. That means a message, a theme, sometimes even friction. If an article doesn't have an opinion, it isn't an article — it's a document.

2. Engagement numbers without quality checks are self-deception. I regularly ask my clients and my team members: Who liked this? Who commented? Where are the shares coming from? If the answer is “I don't know,” then we don't know whether we're doing marketing or just producing numbers.

3. AI optimisation and human-first content are not mutually exclusive. You can write an article that real humans want to read — and still ensure it's technically sound for AI search. The difference lies in the sequence: first, write for the human. Then, optimise for the machine. Not the other way round.

4. Volume is no substitute for strategy. More content does not mean more impact. I've seen clients achieve more with one post a week than with daily AI output — because that single post had a clear message that made someone think.

The real question

I believe we are at a point where the industry has to make a decision — not collectively, but every agency, every marketing department, every content creator for themselves.

Are we optimising for the machine that ranks? Or are we communicating with the human who buys?
It sounds like a rhetorical question. But it isn't. Because the answer right now is anything but self-evident.

I've made my choice. And I'm curious to know where you stand.

Let's talk about it on LinkedIn