Occasionally you find other players, but they often quit fast, since it seems most are just checking the game out and are not interested in playing.
You can join on the #duion.com channel on freenode IRC, to find other people to play with.
Bots are in there in the latest development, but as I have many levels, gamemods and weapons it will take time to teach them to handle it all, currently they only can play deathmatch with regular weapons. Maybe I will just leave it with that and make a release later and host a deathmatch only server with bots.
I was reading a thread recently about an online game that is shutting down where someone made a good point about the challenge online multiplayer-only games face. The premise was that for a casual visitor who likes the idea of the game and wants to support it, there is literally nothing to do when they want to play and nobody else is on the server. So the game is considered boring and eventually a player loses interest and walks away. In the context of the comment thread, the dynamic was referred to as a death-spiral, but from Ubergame's angle the same dynamic seems to be a serious hurdle to establishing a player base in the first place. Looking at the Steam reviews, I'm seeing a reasonably common thread of similar sentiment in the comments - both positive and negative.
If players had something compelling to do on an otherwise empty server, they would probably stick around longer and the chance of getting a random human matchup would seriously increase. If fighting AI can be made even somewhat fun, it might be enough to get a few people coming back regularly for the diversion with the intermittent reinforcement of finding a human matchup acting as a bonus.
My other (unsolicited) feedback is that the levels are way too big for the type of game - especially with low player-counts. It's a deathmatch shooter but the maps lay out more like an open-world adventure. You want players to be fighting, not running around looking for each other.
I did some popping between Ubergame and Black Ops II (which, IMO, is one of the best in the genre) trying to get an idea for what they've done to achieve their feel. The consistent answer was that the maps are all reasonably small, very full and have strong consideration for sightline ranges. Also, the overall visibility range is *much* lower (scaled to play-area size maybe) - so the levels start to get distance-fog basically at the outer edges of the longest-range sight line. They end up being more like mazes than open levels and it's fun as hell.
Ubergame's paintball areas are very nicely set up for small/medium-scale deathmatches, but each of the play areas should probably be closed off, stacked as individual arenas and presented as different levels. Eventually the bridges and stands could be filled with dumb wandering-ai milling toward the active area and it would make a pretty neat effect. It feels like the other levels should be reduced/tightened to varying degrees and then have the sightlines get a final tweak with a bunch of environmental objects. Obviously, I know the amount of work I'm talking about and the fact that you're just one person, so take the level-design feedback for what it is.
I'm kind of on a personal TGE porting mission at the moment ... and need to crank out some "real" code to pay the bills here in a week or so. But one of the things on my list is porting a general AI resource and bringing in whatever innovations have found their way into the community since it was written several years ago. I'll take a look at your codebase and see if maybe I can help out a bit with the AI while technically still working on my porting mission.