Microsoft refuses a documentation fix because AI might not understand text in tables
Microsoft refuses a documentation fix because AI might not understand text in tables
Oh get fucked. This was exactly my fear, as AI use ramps up, the web will quickly be designed to cater to AIs over real humans.
Get double fucked, mattwojo of Microsoft. Take that stupid handlebar mustache and shove it up your ass.
it is chefskiss.tiff that I keep “having conversations”[0] where people tell me that “AI” is “well-suited” to “data extraction that traditional tools struggle with” and then datapoints like this keep. coming. up.
[0] - I…am not inviting these. the promptfans just Defend their twisted little hearts away, unasked. it is tedious.
Wojciakowski took the critiques on board. “Wow, tough crowd … I’ve learned today that you are sensitive to ensuring human readability.”
Christ, what an asshole.
Imagine that.
To be fair, he did ultimately merge the changes. He deserves to be ridiculed for the awful priorities, but a lot of people would dig in.
It’s even worse when you add the next few words:
I’ve learned today that you are sensitive to ensuring human readability over any concerns in regard to AI consumption
The machine readable docs is the docstrings (or XML Documentation Comments or whatev), and the code itself. LLMs have completely melted these people’s brains.
Give AI access to some simple algorithms. For example, an algorithm that takes a table and extracts a column into a comma-separated list of text strings.
This is how humans “comprehend a table as text”: we read across rows, and we read down columns. Values that aren’t in the same row, or in the same column, are less related than values in the same column or same row.