The thing with Generative AI is that the user has to make the right prompt.
These were very broad questions that I asked just to put forward the point on how this can be used for the game.
Don't disagree in principle. There's certainly a lot more data points that when comparing reality and the game doesn't necessarily match up.
Remember doing my Lockdown Tournament on Cricket 19 and there was an awful lot that didn't mirror the real sport. In terms of wickets lost in T20s, a big factor was AI opening partnerships were a lot stronger and more consistent than you'd see from the majority of sides.
Agree with TJ though, there can be too much accuracy where it becomes as predictable. Less about what the AI is doing in response to you and more just doing what it's told. I'd argue that it's tough to find that sweet spot where both applies. That's why I am always in favour of more strategic or tactical gameplay elements.
Other factor is how to make things like par scores apply in a similar way to the human player. If the AI consistently falls apart for less than 200 on a specific pitch, then there needs to be sufficient gameplay challenge to make that a similar likelihood for the human player, imo. If you achieve above a par score then it should feel like an achievement but that only happens if there's struggle to do so - likewise if you fail to keep the AI to a par total it should be clear what you could have done better. Whether it's poor bowling (and what poor bowling qualifies as) or dropping catches.
