It feels like the AI will make both ludicrous choices for no reason, and then later in the game play exquisitely and not blunder once—like it plays at a 300 level here and there, and then 2000 level in the same game. I imagine it must be very difficult to balance AI to feel consistent.
Engines were made to find the best moves possible. It's hard to make engines play at lower rating levels and even harder to get them to play realistically at those levels
The attempt is achieved by decreasing how deep or long they search and changing some engine parameters that will make it play less optimally.
It feels like the AI will make both ludicrous choices for no reason, and then later in the game play exquisitely and not blunder once—like it plays at a 300 level here and there, and then 2000 level in the same game. I imagine it must be very difficult to balance AI to feel consistent.