According with next description (https://www.servicenow.com/community/ai-intelligence-articles/under-the-hood-now-assist-in-ai-search..., AI search sends the top KB articles to LLM to generate an answer.
The question is, the summarization is made from several KBs (the Top) or just one single (the most relevant)
I´ve been doing some tests and LLM and in my test, LLM summarizes one single KB, even when similar KB are found as relevant
For instance, I made two KB explaining (how to do something) split in steps from 1..4 in the first KB and steps from 5 .. 8
AI search find the top KBs both, but LLM only summarizes the first KB found
