Although the funniest thing about the holodeck these days is LLMs have answered a question: can you have realistic non-sentient avatars? Evidently yes, and holodeck authorship is likely a bunch of prompt engineering, with really advanced stuff happening when someone trains a new model or something.

Similarly in Stat Wars with droids: Obi-Wan is right, droids can't think and deserve no real moral consideration because they're just advanced language models in bodies (C3PO insisting on proper protocol because he's a protocol droid is the engineering attempt to keep the LLM on track).