Presently, AI has become profoundly entrenched in the realm of long-form text creation, particularly in academic paper writing. Yet, traditional AI text detection tools are limited to discerning text origins based solely on linguistic characteristics, posing a challenge in pinpointing the source of research ideation within human-AI collaborative efforts. To tackle this issue, researchers from pertinent institutions have crafted the IdeaLens detection system. This system transcends the constraints of language style scrutiny by extracting and eliminating the interference of original phrasing, honing in on content outlines to ascertain the origin of text ideation. Numerous tests have demonstrated the system's exceptional performance in scenarios where distinguishing between AI-generated ideation and human-authored content is crucial. It supports cross-language detection and adeptly differentiates between the ideation traits of humans and AI. At present, the relevant training and evaluation datasets have been released to the public. However, the system still has its limitations, such as training tags that are not directly labeled, restricted recognition capabilities in certain scenarios, and an inability to discern the collaborative ideation contributions of humans and AI.