Leadership CONNECT: 15-MAY-2025 (Y25W21)

Conclusion 1: Defeating the Devil’s Plan

These devastating digital results are here and now (one good starting balanced research is presented in this paper hosted by NIH: The impact of digital technology use on adolescent well-being – PMC).

Awareness is a start, but not enough. We must actively combat this plan.

p.s. If you are a parent, think again and again about giving your child a screen. You are zombifying them.

Conclusion 2: Learning From This Case

 I found both the content about the devil’s blueprint, so AI sure gave me an answer. I found both the content about the devil’s blueprint and its viral version to be highly valuable. Of course, additional editing and fine-tuning (with AI) would have enhanced it even further. For this post, I deliberately kept it almost as is—to demonstrate what can be considered “zero-shot” value. 

Based on three years of heavy AI use, I can state that when the models improve at detecting human intention and humans become clearer in their prompting (especially on dynamic prompting), you achieve better results more quickly. It is a moving target, as tools get better and we know more about how to use them.

Appendix on my AI Stack: Tools I Used Today and Some Surprises (!)

1. I used the automatic “pic to text” feature of Apple IOS to extract the core question from an image I took from the video on LinkedIn.
2. I asked ChatGPT the question. Hoping to generate an answer in the form of a table.
3. ChatGPT did not want to answer, but suggested a way to bypass (!).
4. ChatGPT created the core table and then suggested a visual (actually giving me two options).
5. A day later, I logged back into the same conversation (this memory thing is cool — you continue exactly where you left off).
6. Asked for a cool name for the post.
7. Firmed on the name of the post, after a few iterations. GPT is excellent here.
8. ChaptGPT suggested writing the post for me (!) (Note: ChaptGPT will increasingly suggest one action, not allowing you to think independently).
9. Great post, editing it in Canvas.
10. Editing using Grammarly (which works inside Canvas).
11. In the same chat, I started to generate the cover image (see expanded prompt) — A realistic, hedonistic-looking bluish devil with subtle human traits, lounging or smirking with confidence. The devil should appear stylish yet menacing, symbolizing modern digital temptation. Set against a clean, vibrant orange background. The composition should be high-quality, centered, and suitable as a striking visual for a thought-provoking post.
12. I got the first image.
13. Trying to edit it, Chat suddenly said (!): “I can’t generate that specific image as it violates our content policies. Please feel free to suggest a new prompt or direction, and I’ll be happy to help!”
14. I decided not to continue playing with this because it is already four hours of work (!)—this AI thing does take time. 
15. I moved graphics to Canva to create a cover (5 m).
16. Final editing on Google Docs + Grammarly.
17. Move to production (web, LinkedIn, newsletter).

P.S. Would you like to see more behind the scenes of AI use? Email me.