From a practical perspective, because of inertia and all managers seemingly being "burned once, twice shy" on re-designs.

If my company already uses spreadsheets and teams of people skilled at manipulating them, it might not be very hard to convince my manager that good AI use cases (quick formatting, intelligent auto-completes, intention detection from sheet structure) could probably speed us up by 20% - but convincing management of the value of a full system re-design by comparison would likely be a very tough sell, if not viewed as an insult or sign that I'm not a team player.

Just because the future may prove these methods antiquated doesn't mean management is liable or likely to change their tune today.