I hope this blog post finds you well. The Erasmus+ Staff Training Program empowered me to embark on an exciting opportunity to participate in the AI Summer Lab. The exceptional teachers and the incredible organizing team did a fantastic job with this deep dive into AI that was a truly transformational experience.
I spent the week of 14–18 July at Masaryk University in Brno, thanks to the Erasmus+ Staff Training Program. Officially, I was there for an AI Summer Lab. Unofficially, I was there to overload my brain with new ideas, drink much coffee (one even served by a robot), and try to resist drinking too much Czeck beer.
The teachers were excellent, the organizing team tireless, and the schedule… intense. And because we had just learned to spot “AI-sounding” sentences, I couldn’t resist starting this post with one. (Go on, read the first line again and tell me it doesn’t sound like something ChatGPT would write.) Right after the week ended, my head was still spinning. Now that I’ve processed my notes, instant reflections, and the presentations, here are my key takeaways.
Spotting AI Language in the Wild
Our teacher Katka made it quite clear, that we face a shift in our role when it comes to dealing with texts. We won’t be writers anymore, but proofreaders. For someone like me, who likes writing, not a very compelling outlook.
It will be interesting to see which way the trend goes:
- Will people start speaking and writing in an even flatter, AI-like style?
- Or will AI become more sophisticated and somehow pull human language up with it?
For the latter to happen, we’d need the right training data — and I hope there’s still enough of that left. If not, maybe humans need to upskill in finetuning texts (whether they’re written by humans or AI).
The “Do I Really Need This?” Rule
Quite a clear message I took with me was: “Reflect first, whether what you are going to do with AI is really worth it and necessary, or a waste of resources!”.
It’s dangerously easy to create things that are, at best, mildly amusing but ultimately… garbage. And considering the energy and water required, it’s worth asking: Is this worth it? On the other hand, to learn about AI and how to best apply it, you have to experiment. And not every idea will work out. So, is it possible to build the necessary skills without some waste? I decided to try things at least once, see how they work and to be able to judge what could be possible applications. Take the HeyGen video I linked below — fun to make, yes, but, honestly, utterly useless. But it was fun to make. Nevertheless, I won’t produce any other videos until, one day, I might in fact need to produce a video that actually generates a benefit. So I wasted some resources to build a skill, but refrain from using until a senseful use case pops up.
What AI generated, based on my notes
To practice what we learned, I fed my summarized notes into AI tools to see what they would produce. Here is what I got, based on tools and ideas I learned about during the AI Summer Lab:[1]
- Infograph (Gemini): Link
- Webpages (Gemini): Link1 and Link2
- Podcast (NotebookLM): Link
- Thank-you-video for the team at Masaryk University, highlighting the transformational character of the AI lab (HeyGen based on a Copilot picture): Original and English version
- Slide deck[2]: Link
Final Thought
The AI Summer Lab didn’t just teach me about tools, prompts, and outputs — it left me with a bigger question:
How can we make sure AI helps us write, speak, and think better instead of flattening everything into the same polite-but-empty corporate style?
Until I figure that out, I’ll keep experimenting — but I promise to ask myself first: Do I really need this?
Disclaimer
I wrote the first draft of the text myself. First time in English. I reused the project in ChatGPT with all my previous posts and asked it to make suggestions – this time with a focus on transferring my personal writing style from my native German texts to my English text.[3]
[1] I only made a few adjustments to the webpage (adding images, videos, and a copyright notice, since the first draft from Gemini only included placeholders for images). With the infographic, it was immediately obvious that the uploaded file wasn’t suitable because it contained no data to process. As a result, Gemini simply invented some distributions for bar and pie charts.
[2] It took some circles until Copilot (in my paid M365 family license) processed my notes, but the resulting ppt was way better than the version from ChatGPT. It still needed some editing, but this was definitely a use case where I could see the amount of time saved.
[3] The prompt was: „Hi, I drafted a new article for my blog. Can you proofread and make suggestions on how to improve the article? Please keep in mind and have a special focus on the fact that this is my first article in English and English is not my mother tongue. Can you help me to transfer my style of writing in the old German files to this new article in English?”