One would think that all the books that Google has scanned and OCR'd would give them some kind of edge in the LLM-AI department (maybe a virtual research assistant?)