Even just using it once to detect that my move would be a blunder, or show that the my opponent’s last move was a blunder (that I need to now figure out how to punish) is too much, in my opinion.
The problem is that the SE is available and hence is implicitly condoned by it being available.
Or even less than that - just using it to see if you are ahead and should play safe or behind and you need to do something big to catch up is AI assistance beyond what I would consider fair.
Sure for me too. And using a joseki database too I would say. Honestly.
I’m not for both of them but you know I’m curious of what are the expectations around.
I think the reason josekibases are allowed is because in the olden days, it was generally accepted that people could spend days poring over books and using whatever resources available to make the next move (through mail). Besides, it still takes some thought to determine which joseki to play or whether to deviate. AI kind of throws all that out the window with “objectively correct” (beyond human level) move suggestions and scores.
While we are at it, I’d like to mention one more way to use such an analysis tool to cheat.
Even if the score estimator is disabeled during a live game, people could still open a demo board, put down the position and ask the score estimator for insight.
If you do such a thing on lichess.org, you’ll be detected as cheater and automatically forfeit your game. Such an automatic detection would be nice in my opinion.
At that point you might as well be running KataGo on your own computer. I would consider any demo board with AI evaluation a blatant violation of the ToS.
“Next: Should AI analysis be on by default during games?”
Seriously, I don’t know if I’m missing something but if the discussion/poll is about being able to see Kata scores during games via score estimator (without these estimations being weakened), that would mean being able to check how big a mistake you made after each move, whether you played the AI preferred move or not and also being alerted to gamechanging L&D situations that most of us often completely overlook.
Discussing whether this should be available during games is just one step short of discussing whether we should be able to consult AI during games.
The “behind the curtain” debate will soon be about means each of us can have considering AI (and SE).
And one side you’ll have the people against use of external tools. These are degrading the spirit of rated games and so on.
And the other side you’ll have the people thinking that it’s unfair that some people can run their katago, SE and so on and themselves, they can’t because of the costs so they want to rely on the server to make it more fair.
If it is included, then we should have an option to refuse auto-match for a challenge with the AI score estimator enabled.
I like to just click auto-match to get a live game, I wouldn’t want to get paired with someone that created a challenge with the AI score estimator enabled.
In ongoing ranked games, Kata SE should not be available on analysis branches. If you can explore the expected score for multiple options, you simply end with the AI preferred move.
Territory marks give a strong clue which parts of the board are (un)settlled, so they shouldn’t be available in ranked games at all.
As mentioned the AIs opinion can be interesting in teaching games and there like, so it should be (optional) available for unranked games.
As a spectator I’d like to have the SE + territory marks available, as I’m basically reviewing a ongoing game.
when allowing it for spectators, cheating with the AI SE becomes as easy as opening the game in a private tab… if the owner wants it disabled, it should be disabled for everyone on that game, player, spectator or guest, while the game is being played.
One could have it disabled for guests. As in grey out those options or block them the same way other features like that work.
Then I think someone would have to log in with an alt account to spectate and use it. Bit more work involved for the person trying to do it, but still possible I guess.