GPT-5.5 authorship and order effects
While comparing medium, high, and extra-high reasoning modes on the same task, I uncovered consistent authorship and presentation-order effects in GPT-5.5’s evaluations.
Guillaume Valmont's Weblog
Guillaume Valmont's notes on AI-assisted web development experiments and learnings.