The moment you start playing for ELO instead of fun is the moment you should reconsider playing.

The estimated ELO, I think, is mostly based on accuracy. It may be that it's much easier for you to play accurately against a bot than a more erratic human player.

It's very difficult to write a chess bot that plays like a human. They've improved in recent years in their human-ness but will still make very obvious mistakes that a human player is unlikely to make even at low Elo. This makes them a fairly poor substitute for real players if you want to practise and improve.

Rating anxiety is also common in games that have rankings systems. There are various strategies for avoiding rating anxiety, but you could try setting yourself your own goal such as "I'm going to play 5 games against human players regardless of the outcomes" and try to disconnect your performance and rating from your sense of accomplishment. If you analyse your losses and learn from your mistakes then even if you lose you are gaining something, so you should have less anxiety about losing before you start playing.

The first part of any chess game really comes down to recognizing certain patterns of setup or attack, which with lower level bots will be pretty basic and the same. You can do well against people at least in the first parts simply be expanding your knowledge of know openings and how to deal with them, especially the ones that are designed to trap you into an early checkmate. Often times those are the worst to deal with online until you can see what they're doing and counter them, and many of them will resign to avoid actually getting into a deeper game.

Even if the ELO is different, in theory after many games and enough consistency in your gameplay you should settle out into a level where you're finding mostly equal players. If you do sometime fall lower, again in theory better playing should bring you back eventually.

TF is wrong with me.

Or are those estimated ELO 100% bs?