Some topics just don’t let you go. For me, it’s AI right now. I’m testing models, tinkering with the technical possibilities, and constantly exchanging thoughts with industry colleagues. And I keep running into two camps:

One says AI will replace everything – from planners to creatives. The other and that’s where I now count myself, sees it differently: Yes, AI is changing a lot. But no, it won’t simply replace customer service, procurement, or even the human gut feeling.

Why? Because we’re repeating the same mistakes we already made during digitalization and in digital advertising.

Technology Over Strategy

95 percent of AI pilot projects in German companies fail (t3n, Handelsblatt).
The main reason is alarmingly simple: companies believe they can just layer a few tools on top of existing systems and, voilà, everything works. Spoiler: it doesn’t.

This reminds me of the digitalization wave 10–15 years ago. Back then, we slapped software onto silos and hoped it would magically become an integrated system. It didn’t.

The GIGO Principle – Garbage In, Garbage Out

With AI, the old rule “Garbage In, Garbage Out” hits even harder. 70–80 percent of projects fail due to poor data quality (AP-Verlag,Tale of Data).
42 percent of companies even report that more than half of their AI projects were delayed or failed because of data issues (Akaike).

Bias, hallucinations, and large-scale wrong decisions – all of this happens when the foundation isn’t right (Vodworks, Integreon).

From digitalization mistakes to AI mistakes: the industry’s déjà-vus

Tool-stacking without a plan: Companies pile up AI tools, but processes and organization remain the same. The result: many small islands without real added value.

Data in silos: Less than a third of companies are prepared for the data requirements of AI (Heise).

Forgetting change management: Only 32 percent of employees feel confident using AI tools.

Extra Challenges With AI

Maintenance: AI systems need constant re-training – that costs money and is underestimated.

Blind trust: Many simply believe AI results. But they’re only as good as the data you feed in.

Lack of expertise: Only 40 percent of companies have enough specialists to properly run AI (Zentrum Ilmenau).

The Price of Bad Data

Poor data quality costs companies an average of $12.9 million per year.
Unity Software once lost $110 million due to a single corrupted file (Vodworks).

My Takeaway

With AI, we’re at a similar point as we were with digitalization. Lots of hype, lots of tools – but the basics are missing: data quality, clear processes, change management.

The big question is: have we learned from past mistakes? Or are we heading straight into the same wall again?

I’m optimistic. But only if we take the fundamentals seriously – instead of hoping technology will magically fix everything.