Deploying AI into Chessbase

[bmp1974]

It is high time chessbase thought about implementing AI tools to enhance its capabilities in its next release.
Some areas could be to improve annotations of games, giving historical information when going through a game, opening history or tips, anecdotes, player history or interesting facts about players etc.

Is there any information if such AI integration into chess is already happening?

[CornfedForever]

It is high time chessbase thought about implementing AI tools to enhance its capabilities in its next release.
Some areas could be to improve annotations of games, giving historical information when going through a game, opening history or tips, anecdotes, player history or interesting facts about players etc.

Is there any information if such AI integration into chess is already happening?

I feel sure that hast to be in the cards.

That said...most of what you just mention would largely be worthless for me and I use my Chessbase many times a day analyzing games. Some professional Chessbase critics would of course likely find more chances to shout 'bloat' or some such. But if you do it right...it could be worthwhile.

Improved 'auto-annotations) would always be good. With my new laptop I compared the annotations/note to a blitz game I played the night before on my 'experimental openings account', letting 'Tactical Analysis' on the less cluttered ' Coarse' option at the recommended 3 sec/move for both Lc0 and the current Stockfish Development version. Nothing serious of course and I probably should try more time and deeper notation (like Fine).

At this setting, sometimes the engine output was exactly the same....but with different words used for the notes. Sometimes a line was pointed out by one, but not mentioned by the other. Once a '?" was offered to a move I played and there was nothing wrong with it...not even sure the alternative by Lc0 was any better and the Lc0 offering seemed to make less sense than my move...and if I just 'go to the position' and analyze for a wee bit outside the 'auto-analysis' feature, each seem to think my move is really no better nor worse than Lc0's and not deserving of a '?' Stockfish's comment is to point out what the idea indeed was. On the follow up, Stockfish pointed out the sacrifice I was considering was ineed correct but a large margin...Lc0 remained silent. Me...I chickened out.

Stockfish clearly offered more variations/evaluations.
Weighted error evaluation was about the same proportionally with Stockfish giving lower scores.

I took a snippet of each...if I knew how to post it here I would.

Oh and that '?' comment brings up a favorite bugaboo of mine in regards to ANY auto-analysis feature I've seen online or even Chessbase. If you do not want to introduce new evaluations...at least color code the symbols. A GREEN '?" should mean a perfectly good move (!?) would be better if it does not truly offer a chance for the opposition to substantially change the course of the game...while a RED "?" should indicate an actually poor or bad move.clearly divided into OPENING, MIDDLEGAME AND ENDGAME with 'weighted errors' broken out by each. They already point out combinational opportunities utilized/missed and such...so being able to slot them in under each phase of the game would be nice. Don't grade me on my 'crap opening experiments'...I'm just wanting to get them out of book.

THEN...being able to consider a set number of game and export that info to, lets say Excel, to get a report indicating weaknesses (think an improved version of Aimchess) to work on would be a 'game changer' as they say.

I am not sure that fits neatly into your AI bucket...but it's where they need to go first IMHO. So much still on the table.