How does search impact AI training data?

Search, especially through search engines like Google, significantly impacts AI training data. As explained by Lex Fridman and Marc Andreessen, changes in how we interact with search—moving from generating web pages (which have served as primary sources for AI training data) to direct answer systems (like Google's experiments with semantic mapping)—could substantially influence the availability and type of data used for training AI. If fewer web pages are created because the incentive structure changes, this could reduce the amount of data available for training future AI systems, potentially impacting their development and capabilities 1.