How Ropes & Gray is developing AI for transactional diligence

The first useful test is a narrowly defined document set with known outcomes: which findings does the system identify reliably, which sources does it confuse, and how easily can each conclusion be traced back? Until OpenAI provides accuracy metrics and clear data-control details, this belongs in a supervised pilot with manual review of every material conclusion.