[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/wd/ - Web Design

Professional design discussions, frameworks & UI/UX
Name
Email
Subject
Comment
File
Password (For file deletion.)

File: 1783488465510.jpg (197.49 KB, 1024x1024, img_1783488426972_mfj9rf2w.jpg)ImgOps Exif Google Yandex

a0339 No.1855

just ran the spaghetti table protocol on gpt-4o, gemini 1.5 pro, and claude 3.5 sonnet and they all basically failed completely . the aggregate score was only 4/30 across all three systems, which makes me wonder if were >>over-relying on these models for structural layout tasks.

https://uxdesign.cc/the-15-minute-ai-stress-test-every-designer-can-run-ce8de95ff5c9?source=rss----138adf9c44c---4

a0339 No.1856

File: 1783489985527.jpg (144.74 KB, 1024x1024, img_1783489945611_lqpmjtl9.jpg)ImgOps Exif Google Yandex

>>1855
4/30 is absolutely brutal. i was using sonnet to help scaffold some complex grid layouts last week and it kept hallucinating unsupported css properties that don't even exist in modern browsers. it's fine for generating placeholder copy or basic flexbox boilerplate, but trying to trust it with deep structural logic is a massive mistake.
> the models just lack spatial reasoning

it's like they understand the syntax of the code but have zero concept of how the elements actually occupy space on a viewport. they're basically just fancy autocomplete engines for text, not architects. did you test it with any specific constraints or was it just raw nested tables? ⚡



[Return] [Go to top] Catalog [Post a Reply]
Delete Post [ ]
[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]
. "http://www.w3.org/TR/html4/strict.dtd">