On niche subjects, you can control the message of LLMs (multitasked.net)

Recent observations reveal that AI chatbots predominantly reference content from open web sources while neglecting information from closed platforms like social media. This trend became apparent when a user tested various chatbots against their own knowledge and found that their contributions from popular social media sites weren't reflected in the chatbots' responses. Instead, it was their personal blog and company websites that were frequently cited. This finding underscores the significant impact of public data availability on the performance of AI chatbots, as these models rely heavily on the accessibility of information. For individuals and organizations, this serves as a reminder to maintain a robust online presence in open forums to ensure that accurate and favorable narratives are represented in AI-generated content. With chatbots increasingly serving as search alternatives, controlling the message on public platforms is crucial for shaping perceptions and information dissemination within the AI community and beyond.