The main reason reader mode is "underused" is that for a substantial fraction of pages, it does not work.
Because of the way the human motivational architecture works, if there were a reader mode that always worked, its rate of usage would be many times higher than the usage rates of the current crop of reader modes.
But of course, there is no way to create a reader mode that works for all web pages -- and that fact is one of the main arguments for competing with the web rather than trying to improve or fix the web.
By "competing with the web", I mean creating online services that take the attention of users away from the web for some subset of things the web is used for, e.g., consuming static textual content, also known as reading.
Gemini would be an example of an attempt to compete with the web, and it has the property that every page is essentially automatically in reader mode.
The trick is to find some desire that the web is currently bad at satisfying, then improving Gemini to cater to that desire. For example, TOR was created (many years ago) by the US military to give its employees a way to browse the web without revealing their browsing history to the spy agencies of the US's enemies. But maybe Gemini's much greater simplicity would enable it to make a stronger guarantee of user privacy than the web is able to guarantee. (The more complicated a web browser gets, the harder it becomes to make privacy or security guarantees -- particularly guarantees of interest to only a small fraction of the web's users.) If that were the case, then whoever needs that stronger guarantee would tend to become avid users and proponents of Gemini. Getting very small numbers of avid users is considered an effective way to start increasing the number of users of a new online service.
The idea I just described is probably a bad idea: there are probably other desires that Gemini or some other non-web service could cater to that are much more effective ways of creating avid users of that non-web service. (For example, the idea I just described has the disadvantage that even if the proposed improved version of Gemini could make stronger privacy guarantees than the web can (which seems unlikely to me) good luck convincing the average user who is not a security professional of that fact.)
Finding them is hard! But trying to find one is a more potent way to try to improve things IMO than trying to fix or improve the web is.