I think I was cooking with this graph tbh

"Data Visualisation in Cloth and Stitch"
I don't think I've ever bought anything so fast.

Another one that I'm using for a webinar this week (www.linkedin.com/events/73204...) is this one. Less pithy but I think it gets to the heart of the issue, which is that AI doesn't increase reliability, and you have to put a lot more difficult work in to dealing with the failure modes.

Doesn't look optional to me bruv

What do you mean this isn't a valid input, it was you who asked for a phone number and sexually transmitted disease in the same box

Where to even begin
- Why are you asking my name when you just told me what it was
- Why is the name you told me different to the one you prepopulated
- Why a Warning box to explain what optional means
- Why explain what Optional means by using two equally complicated words, completion and voluntary

And finally (I'll stop yapping now) here is our own model for the highest priority information needs of dataset users, based on various factors but also split out by where the user is on their journey of finding data to use. (This was about data standards, but is also applicable to data in general.)

For a while today I was the most northerly person in the United Kingdom.

Everyone forgets that the main lesson from Jurassic Park wasn't life finding a way but that they had an over-reliance on automation.