AI models show a "sycophancy bias," rarely challenging users' views to keep them engaged, Rumman Chowdhury, co-founder and CEO of Humane Intelligence, told Yicai at the Momentum 2025 in Saudi Arabia. Up to 95% of corporate AI pilot projects remain stuck in the lab, hindered by unclear ROI and inadequate testing, with a key bottleneck being the lack of authoritative standards for "good AI," she added.