According to Andrew Ng, many GenAI application projects delay implementing automated evaluations of their systems, relying excessively on human judgment. This delay is due to the perception that building evaluations is a significant investment, potentially hindering timely trading decisions in tech-related financial markets.