Hey, it’s Marek.
I’m back after a few weeks away. I lost my dad, and I needed to step offline: spend time with family, disconnect from everything that wasn’t essential. Some of you reached out. Thank you for that.
This episode is about my thoughts on trying to talk to AI during those first days. I asked Claude if I should push through and keep writing, and then I found myself staring at its scratchpad: watching it reason through how to respond to my grief. That’s when it hit me: I needed an answer from the heart, not a calculated response.
The Pattern: Unreasonable Moments
The better AI gets at reasoning, the worse it becomes at unreasonable moments: times when being present matters more than being right. When you need to be heard, not given a solution. When you need to connect, not optimise.
It’s like someone taking notes at a funeral or optimising wedding vows. Technically fine. Also completely missing the point.
I thought about ELIZA, that ancient 1960s chatbot I ran on my BBS in the ‘90s to chat with users when I wasn’t around. No reasoning, no intelligence. Just reflection. People loved it because it witnessed without trying to fix.
Not everyone needs GenAI in sensitive moments. Some people do: no judgment, no pity face, no awkwardness. Dave Hughes showed how he uses it to cope with grief, and that’s genuinely powerful for some. But I needed to recognise my moment for what it was: unreasonable.
Thanks for waiting. Good to be back.
Stay curious!
Listen to the full episode where I share a short story about my dad, my first failed software customer, and what happened when I tried to delegate grief to a reasoning engine.
