Which is all kinds of depressing and horrifying.
The AI behavior in ED, especially of infantry NPCs, is abysmal...but I can't think of many multi-player games with significantly better bots either.
Of course, I'm judging the AI against myself. This may seem unfair, but it's ultimately the baseline any AI is going to need to hit to feel immersive, rather than cheap--and is why I've heavily preferred multi-player titles whenever practical. Of course, most real people are strategically and tactically incompetent as well, but there is enough of a spread, enough exceptions to the rule, and enough surprises from the otherwise mediocre, that they never really get old. EDO's NPCs got old the second time I played the suit tutorial.
Well, yes. There are some problems with multiplayer and AI. It's difficult already to design AI that's advanced enough at decision-making to be an entertaining(!) match for players in a single player game, and add to that that multiplayer games could even have a wider range (in terms of skill) of players playing them. As you yourself noted, you are judging the AI against yourself: one that offers an entertaining challenge to one person could easily turn three others away. So it's a difficult act to balance.
Anyway, I believe most multiplayer games these days don't feature advanced bots because they don't see a need to: the people who would challenge themselves against those would be just as likely to go fight against other players. So why spend time on making good AI if the majority of people wouldn't enjoy playing against those?
Or, put another way: suppose we had possible bot skill choices go from 1 to 5, and the majority of players would choose 3 anyway. Then if a developer were to come in and would add better AI on top, extending it to 1 to 10, only those people would appreciate it who found 5 to be too weak for their tastes already. For the majority who enjoys 3, they might not like it if they are told that their level of play equals 3 out of 10 instead of 3 out of 5. The work that was done ended up improving the experience of only a smaller subset of players, and slightly worsening it for the largest subset. That's a scenario developers would generally want to avoid.
If memory serves, a similar problem happened somewhat early on in ED too. Prompted by frequent player complaints about how terrible NPC ships behaved in station traffic, a developer was tasked with improving their flight, and after successfully doing that, she was directed to work on combat too. That went on for some time, but in the end, enemy ships had gotten too good for too many people that the AI was dumbed down from its best. After all, even then, players could encounter Elite-ranked NPCs easily enough: with today's fast rank progressions, it would be even more so.
What I don’t have is much to choose from when I look for a Space Sim MMO. There’s Elite Dangerous, two failed attempts making a PvP one, one that toes the line between incompetence and a scam, and two outright NFT scams. Those are some patheticly paltry pickings for a genre of MMO that I would’ve thought would be been pretty popular. Why isn’t there a Star Wars version, for example? Or one set in the Expanse? How About Traveler Online? One set in the Battletech universe?
I think a lot of it could well have to do with the fact that in a space sim MMO, clients have to go with full six degrees of freedom. That can quickly add a lot of complexity to the code, you can't make some handy optimization tricks that you normally could. After all, in your average MMORPG, the only coordinates you usually need for characters are them running around on x and z of the terrain, changing their y height above the terrain if they press jump, and usually, what angle they are looking at. (For the purposes of enemy players or NPCs backstabbing the character.) So that's a lot less to consider and communicate between clients and servers than what you get with six that change all the time. Obviously the latter is not impossible, but it is certainly more costly, and you have to factor that in. For making a new MMO game, the expected costs are going to be crucial.
Add to this that introducing realistic scales complicates things a lot more: practically, you have to design your engine to do this. See: Star Citizen trying to shoehorn larger scales (yet still not realistic 1:1) into a licensed engine and still having spectacular bugs after so many years. But then, No Man's Sky run their own engine, and they still opted for lower scales too.
Oh, and another thing: cosmetics. I think that most people assumed that cosmetics for human(oid) characters would sell much better than cosmetics for spaceships would. However, after Star Citizen, I wouldn't be so sure about this

But anyway, it's a tough pitch.
Then there's the history. Elite: Dangerous didn't blow up the genre, Star Citizen has... a reputation, and later on, the game that was hyped to blow up space games, No Man's Sky, ended up getting known for its disastrous launch instead. With such prior examples, is it any wonder that there's no line of developers all trying to make the next big space MMO?