2 articles.
The most-cited study of AI grading put 80 multi-turn questions to GPT-4 and 58 expert annotators. GPT-4 matched human preference 85% of the time, above the 81% two humans managed. But when researchers took an answer that was already correct, added a redundant restatement of it at the top, and re-judged, Claude and GPT-3.5 were fooled 91% of the time, every judge favoured whichever answer appeared first, and on maths a reference answer cut the judge's failure rate from 70% to 15% while plain step-by-step reasoning left 6 failures in 20.
Local Law 144 obliges employers to run an independent bias audit on automated hiring tools, publish it, and notify candidates. It says nothing about fixing what the audit finds. A 2025 FAccT paper analysed all 116 publicly available audits and found most do not report results suggesting a violation of the four-fifths rule, while also showing the underlying tools could often be in violation once missing demographic data is accounted for. HireVue's own blog says whether its tools qualify as AEDTs is the employer's call. Here is where the enforcement pressure is heading.