The blog post is doubly bad because any "failures" involving images and image understanding can't necessarily be traced back to structured generation at all!!!
learningmore 12/22/2025||
“However, with the same model, if you just use the completions API and then parse the output, it will return the correct quantity…”
observecauti 12/21/2025||
[dead]
mikert89 12/21/2025|
please i cant take anymore anti ai hot takes.
emp17344 12/21/2025||
Sounds like a you problem. I’m all for people investigating the boundaries of model capability - if you take that as a personal attack, you’re going to have a bad time over the next few years.
Leynos 12/21/2025|||
While I suspect "hot take" is apt, this wasn't exactly anti-AI. Rather, the author is advocating for their particular way of doing genAI output parsing (as opposed to constrained decoding structured output, they advocate unconstrained decoding with a permissive parsing framework)
swiftcoder 12/21/2025||
How is this an "anti AI hot take"? It's discussing using one type of LLM output versus another...