After I play a game and lose, which all the time, i get this other screen that shows the last board placement of stones and has all these numbers and things i have no idea what they are how iam supposed to interpret these crasy marking on the board — iam new only starting four weeks playing arethey suppose to help me or something if they do how does one understand these triangles and circles inside my and my oppoents stones, on the last move shown before i usally give up ??
You mean the AI analysis? It shows what the bot think is the “best” continuation, but take it with a grain on salt. If you resign in a losing position, the “best” moves can be somewhat nonsense because nothing really helps ://
Its way more useful if you check some earlier position where both players still had a good chance of winning the game ^^
2 Likes
It can be distracting, though, I agee, and for a beginner it really dos not mean anything.
@NEWOLDGUY you can turn that off by clicking the “disable AI” button on the right.
I recommend that you do disable it.
I’m inclined to make “disabled” the default for beginners.
Would that cause outrage? 
7 Likes
thank you for explanation i will do that
2 Likes
I’d hope not 
But I’ll come and defend 
2 Likes
do you have to do that for every game ? i did unclick the box for that and they still come after the game
@GreenAsJade FWIW, I’m certainly still a newbie with Go, but I find the AI-generated info one of the most valuable ways to be learning.
I evaluate the best move before I make it, then use the AI generated analysis to see how effective (or stupid!) my own choice was. My learning would be much slower without it. 
Idk about outrage, but AI review is one of the selling points of OGS subscription. Might have an adverse impact on revenue
2 Likes
Noob here. How bad is the automatic AI review? I feel like I get some good advice, but sometimes it doesn’t make sense.
I would say it’s not bad at all in terms of the strength of the AI.
The “bad” part is that it’s hard to use an AI graph if you don’t know how to interpret the numbers and markings.
2 Likes
honestly, disabled by default is how I use it. I turn it off after the game before it starts populating the board with suggestions, review the game myself, and then switch it back on to review my review (if I’m taking the game seriously)
2 Likes
You mean the SE, the Score Estimator, right?
That’s something different than the AI review after the game, and thus off-topic, therefore I’ll put the rest of my reply in hide tags.
Click this.
I also do that sometimes, but note these three points:
-
The SE is dumbed down quite some during the game, for fairness.
-
When you click several times, you get different estimates (for different lines of play), I think it can be ~7–8 different estimates you can click through.
-
You can help the SE in giving a somewhat more accurate estimate if you play a few moves in analysis mode that close territory borders generally removing all ambiguous situations where a move can greatly change the outcome.
1 Like
This is an important clarification.
Theoretically, you can’t be using the AI generated analysis during play, which is how it reads.
2 Likes
@trohde @GreenAsJade My apologies if my terminology is off the mark and confuses matters.
If interested, here’s what I do…
I use (mostly) BadukAI’s “Ponder” and “Analyze All” functions to provide game outcome point estimates and/or win %'s after completion of a game played against the AI bot. (I constraint the model’s evaluation time and depth to varying degrees to make for what I feel is a reasonably competitive/challenging bot without my getting crushed all the time 
)
By evaluating my actual moves against the ones deemed superior (per those metrics) for the many alternative open intersections (according to which AI model chosen, like KataGo, KataGo Human SL, or many another uploaded networks), I can either (1) confirm my good choices (becoming less rare these days) or (2) see how to correct my poor choices (oodles of which remain).
Also, after a game is over and the assessment is complete, if any particular poor choice was especially devastating for my position, I start a fresh branch from that point forward using an improved move - and then take it from there to see how things might have turned out.
I then re-run the “pondering analyses” on this new branch… and continue to learn as I go.
Rinse and repeat. Eventually moving on to start over.
To my mind, I thought of this as leveraging AI technology to improve my strategic and tactical skill sets. But if I’m using the wrong terminology or in a way that doesn’t properly communicate to others what I’m doing, I’m sorry to have mixed up my descriptions.
The score estimation (SE) is produced by the AI models, no? If so, since the AI models aren’t speaking to me in my native tongue, I must try to learn to speak in the model’s language. The way these models currently communicate is not through nice, gentle LLM guidance and encouragement - but through raw blunt mathematical metrics.
FWIW, with BadukAI, yes, one can run the model’s evaluation in real time during live play. Although, admittedly this takes some of the buzz out of the game.
Regardless, these metrics are compelling and rich enough for me to gain insights to help me improve. What am I missing?
I think the only confusion you created is that it sounded like you were doing this each move, as you played it. Which would be cheating.
Doing it after the game is fine.
In the context of the discussion, though, you are really far from a “new beginner”.
You are taking a systematic approach, and have already learned what AI does and what it means.
So presumably if OGS AI analysis were turned off when you started, you would turn it on deliberately and use your knowledge to use it that way.
This is far different to a newcomer like the OP, who is really fresh, and confused seeing AI markers on the board after the game.
The time for beginners to turn AI on is when they know what they are turning on, IMO.
(Whether in fact looking at what the AI suggests is an anyway meaningful for beginners is another topic. Since beginners are typically on 9x9, I suspect AI suggestions might be more fathomable. On 19x19 most of the time the approach you’ve described leads to incomprehensible rat holes 
)
2 Likes
You’re too kind. Thank you. I’ve started learning Go about 4 weeks ago, so I sure feel like a beginner
Since I’m playing alone against a bot, I certainly don’t feel like a cheater. I’m using model algorithms to learn - not win. Same as all those years of learning chess with a library of chess books at hand, but with fewer paper cuts thanks to technology.
Of course if I were playing with another human, then using anything other than one’s own accumulated skill and instincts would indeed certainly be cheating… and be of no genuine fun or enjoyment for anyone involved. 
If you’re doing this unranked, that’s perfectly fine.
If you’re playing ranked against bots you’re still not allowed outside assistance. The reason is that your rank would no longer represent your skill, after a few of these.
2 Likes
Makes good sense.
No, I have not played ranked games on OGS or elsewhere. I don’t feel quite ready… perhaps some day soon. But at that point, I fully agree… unless I’m playing au naturel, it would not just be cheating, it would be a pointless waste of my valuable time and energy.
2 Likes
Can you share a picture of what you see and don’t understand?
Depending on your browser settings it might forget that you deactivated the AI review, but not if you play two games in a row without restarting the browser.
