it has to have a list of all IRL names, including rare names like Sirius, and not blank out Sirius when Sirius the star is mentioned, OR when fictional names are referenced
@NoaNabeshima yes easily, and it has to have a list of all IRL names, including rare names like Sirius, and not blank out Sirius when Sirius the star is mentioned, OR when fictional names are referenced
Not significantly more than it already does.
For a task like this you would define a performance criteria, like 90% of names eliminated. The percentage would increase as you add hardware.
Given the stagnant state of the data center market it would probably be years before performance doubled from what it is now.
@MarkIngraham I also want to add the idea of applying linear time to these problems is a fallacy. Technology doesn't improve over time, it's a function of the resources being available when you need them or the tasks being possible at all. It could be possible in 2024 but not 2025, depending if the company fails, there's blackouts, etc.