I've read several dozen articles about the current spate of generative AI, in a context for a client. My thought it, as previous posters have suggested: If asked a LMM (like ChatGPT) to ingest the complete works of SLC, including the Speeches, Letters and Autobiography, in theory you could have something that would respond to queries in a simulacrum of SLC's public voice. , I would definitely add dates to the algorithm in there. 1865 to 1909 is a long span. Assuming you had done this, what would you have? A high-tech Twaint. On the other hand-- suppose you applied your new Twainbot to filling in the blanks on an unfinished MS like Huck & Tom Among the Indians, or #44? You wouldn't have Twain, but you would have something derived from Twain. DDD