DAE notice people blame the chatbot when the real problem is their prompts?
I run a little Slack bot for my team at work and everyone kept saying the model was broken last week. Turns out 8 of the 11 bad answers came from prompts that were just one vague sentence with no context at all. I added a simple template with 3 examples and the complaint emails basically vanished overnight. It bugs me because everyone wants the shiny new model instead of fixing how they ask. Anyone else seen this happen with their team?
Seriously? Eight out of eleven bad answers from vague one-liners sounds less like a prompt problem and more like people not knowing what they want. A three-example template fixing it overnight tells me the bot was fine and the asks were lazy, sure, but is that really a crisis worth being bugged about? Half the time when someone types "make this better" they genuinely don't know what better means yet, and no template solves that. Feels like a small win dressed up as a big lesson about human nature.