Introducing SensAI Go

General conversations about Go belong here.
golem7
Dies with sente
Posts: 102
Joined: Wed Mar 09, 2011 6:51 pm
Rank: kgs 2d
GD Posts: 0
Has thanked: 13 times
Been thanked: 33 times

Re: Introducing SensAI Go

Post by golem7 »

kvasir wrote: It is what the education professionals in my Go club tell me. I don't think it needs much discussion that if you interrupt an exercise that it is likely to have a negative effect. As far as I see your suggestion it is interrupting the task, it is a distraction from the task, but with the nuance that there is some information being conveyed that is useful for completing the task. Yet the question is not if the task can be performed better, the question is if something is learned faster this way. I'm not sure what that something would be, is it the feedback itself?
In your video you show a game in progress on an online platform and some audio feedback at the same time. Now you talk about your experience playing while having such feedback. Who did you play?
I'm not suggesting to interrupt the game for feedback, it is really generated live and runs in the background while I play, maybe that wasn't clear?
I've played various games online against random opponents while I had my feedback program running.

What can be learned?
The main idea is to improve evaluation/judgement, by always being instantly aware of the result of moves/sequences (did the position improve or worsen?)
golem7
Dies with sente
Posts: 102
Joined: Wed Mar 09, 2011 6:51 pm
Rank: kgs 2d
GD Posts: 0
Has thanked: 13 times
Been thanked: 33 times

Re: Introducing SensAI Go

Post by golem7 »

Javaness2 wrote:It's an interesting idea, but I'm not sure that's the best medium with which to deliver feedback during the same.
I think if I was playing I would just find the additional noise offputting. Did you try the same idea but with a purely visual feedback for the key ideas, urgency & state.
I added the visual feedback only later to the program, I originally wanted it to be only audio in order to not distract the eyes from the game. I haven't tried visual only.
In my vision of an ideal future go server that offers such a feedback training mode, each user would be able to choose which feedback type they prefer:
heartbeat audio, visual, music, nature sounds, arcade-style, electrocute when blundering, many things are possible :)
golem7
Dies with sente
Posts: 102
Joined: Wed Mar 09, 2011 6:51 pm
Rank: kgs 2d
GD Posts: 0
Has thanked: 13 times
Been thanked: 33 times

Re: Introducing SensAI Go

Post by golem7 »

John, this Nakane-AI series, can we watch this somewhere? (ideally with subtitles)
dust wrote: It'll be interesting to see this theory tested.

It's also good to have a new idea to shake up the rather familiar discussions on L19 on 'how to improve' featuring a handful of ageing regulars - each with their own idiosyncratic strong opinions, and none of whom are probably improving significantly :) (I include myself fully in that description)
Many thanks dust! :)

FYI: Episode 2 is online:
https://www.youtube.com/watch?v=69Orf_tzQKU
John Fairbairn
Oza
Posts: 3724
Joined: Wed Apr 21, 2010 3:09 am
Has thanked: 20 times
Been thanked: 4672 times

Re: Introducing SensAI Go

Post by John Fairbairn »

John, this Nakane-AI series, can we watch this somewhere? (ideally with subtitles)
No idea. I read it in Go World. It's planned to run for a year so I expect they are aiming at a future book.
John Fairbairn
Oza
Posts: 3724
Joined: Wed Apr 21, 2010 3:09 am
Has thanked: 20 times
Been thanked: 4672 times

Re: Introducing SensAI Go

Post by John Fairbairn »

'Rhythm' came up in this thread, and intuition (or whatever you call internalising information) comes up in lots of threads, and I have a particular hobby-horse in talking about the benefits of studying suji over shape, or the dynamic over static.

It was therefore of great interest to read something last night on this topic but with a different sidelight. There was also a piece of information new to me which I have bolded here, but which was not emphasised in the original. I will give a longish quote so as to provide sufficient context. The overall context is a book called The Everyday Dancer by Deborah Bull, who was a ballet equivalent of a 9-dan go pro, being a Principal with the Royal Ballet (she is now Baroness BUll CBE). I recommend the book strongly. I will assume the readers here can make the relevant connections with amateur go and previous threads.
To an outsider, the first rehearsals of a ballet would probably appear to be rather slow and inconsequential. fractured movement phrases interrupted by some head scratching and sections repeated over and over again as we struggle to claw back the physical sensation of choreography that has not been danced - or even thought about - for months, if not years. It's hard - even for the dancers involved - to see the connection between these early studio calls and the polished performance of the first night.

Some dancers are better than others at recalling once-familiar steps from dusty memory stores. Frequently we remember, but not quite accurately, swearing we did a step on the left leg when the evidence (a video, for instance) proves it was on the right. When ballets are long out of the repertoire, dancers are likely to remember the rhythm and dynamics of the choreography rather than the exact steps, certain that it went 'di dum, dum, dum, and three and four', but unsure exactly what it was. Interrogating the memory too consciously can cause it to retreat for ever into some distant corner of the brain, and too many questions usually prove fatal. ('No, no, don't ask, just let me do it for you.') Even now, as I try to recall snippets of choreography from my dancing years, I know that I'm best placed to do it as I'm falling asleep at night, when my conscious brain is a bit less 'on duty' than it is during the day.

WE talk a lot about getting things 'into our bodies' and it's true that when we really know a ballet, it seems to be stored directly in our muscles and not in our brains. But while 'muscle memory' is a good way of describing how it feels to dance a well-rehearsed role, it's not accurate. The body doesn't function like a memory-foam mattress, storing shapes and movements as indentations within its lean tissue. What 'muscle memory' probably means is that the steps have been repeated so often, and over such a long period of time, that the skill has become unconscious, their 'memory' moved from the centre of the brain to the cerebellum. When you dance these highly familiar steps, it feels as if your body is doing them on its own, without any reference to the conscious brain, as if you're dancing on autopilot. This is a vital stage in progressing from fumbling first rehearsal to expert performance, and it leaves the frontal cortex free for the important business of interpretation and interaction with the other artists on the stage.
golem7
Dies with sente
Posts: 102
Joined: Wed Mar 09, 2011 6:51 pm
Rank: kgs 2d
GD Posts: 0
Has thanked: 13 times
Been thanked: 33 times

Re: Introducing SensAI Go

Post by golem7 »

New Episode online:
https://youtu.be/YNvDvtjqC70?si=3pBJwIdFuqBq8QFI

This one is more back-and-forth than the first ones that were quite one sided, so you can see more action in the feedback system.
Enjoy :)
kvasir
Lives in sente
Posts: 1040
Joined: Sat Jul 28, 2012 12:29 am
Rank: panda 5 dan
GD Posts: 0
IGS: kvasir
Has thanked: 25 times
Been thanked: 187 times

Re: Introducing SensAI Go

Post by kvasir »

golem7 wrote:I've played various games online against random opponents while I had my feedback program running.
That is called cheating.
golem7
Dies with sente
Posts: 102
Joined: Wed Mar 09, 2011 6:51 pm
Rank: kgs 2d
GD Posts: 0
Has thanked: 13 times
Been thanked: 33 times

Re: Introducing SensAI Go

Post by golem7 »

kvasir wrote:That is called cheating.
You're entitled to your opinion. Of course it's an advantage to be aware of the evaluation. But you still have to think for yourself which moves to play.

Anyway, I think I've explained my reasoning very clearly in my article. The vision is to implement the system in a way so that both opponents receive the same feedback to ensure fairness.
I'm merely demonstrating the concept.

By the way, there is a new episode.
https://youtu.be/fkfq3b8qf4c

And since some people commented that they don't like the audio feedback, I have created a 2nd version without the heartbeat sound:
https://youtu.be/1NkRrfEJjgA

Enjoy :)
RobertJasiek
Judan
Posts: 6272
Joined: Tue Apr 27, 2010 8:54 pm
GD Posts: 0
Been thanked: 797 times
Contact:

Re: Introducing SensAI Go

Post by RobertJasiek »

kvasir wrote:
golem7 wrote:I've played various games online against random opponents while I had my feedback program running.
That is called cheating.
It is cheating
- in tournament games unless explicitly allowed by the tournament rules of special, AI-use-friendly tournaments, otherwise
- in rated games regardless of any opponent's agreement because of affecting third players, otherwise
- unless the opponent agrees in advance.

There are study purposes for using AI during games but they ought to avoid any cheating as above.
kvasir
Lives in sente
Posts: 1040
Joined: Sat Jul 28, 2012 12:29 am
Rank: panda 5 dan
GD Posts: 0
IGS: kvasir
Has thanked: 25 times
Been thanked: 187 times

Re: Introducing SensAI Go

Post by kvasir »

golem7 wrote:
kvasir wrote:That is called cheating.
You're entitled to your opinion. Of course it's an advantage to be aware of the evaluation. But you still have to think for yourself which moves to play.

Anyway, I think I've explained my reasoning very clearly in my article. The vision is to implement the system in a way so that both opponents receive the same feedback to ensure fairness.
I'm merely demonstrating the concept.

By the way, there is a new episode.
https://youtu.be/fkfq3b8qf4c

And since some people commented that they don't like the audio feedback, I have created a 2nd version without the heartbeat sound:
https://youtu.be/1NkRrfEJjgA

Enjoy :)
One more video of cheating. It is sickening.

Obviously, wanting to build a system to play a game never compelled anyone to cheat. Cheating is deceit and crookery. It is incredible to think, because one has some lofty goal and desire, that the means toward the end need to be deceit and crookery. Honest people always seem to find reasons to not stoop so low. If only cheaters could be as creative with reasons for not cheating as they can be with reasons for cheating.

When someone cheats in Go it stops being Go. They stop being a Go player and become a cheater. No one should care what a cheater thinks about studying Go. It is like a cannibal chef's ideas about gastronomy or an infanticidal nurse's opinion on postnatal care. It is something so grisly that it evokes an entirely different kind of interest. Such people may masquerade as chefs and nurses but they are not chefs and they are not nurses. Being a chef requires integrity when it comes to preparing food and being a nurse requires integrity when it comes to care of patients. Being a Go player requires, above all else, enough integrity to know not to cheat in Go.
User avatar
jlt
Gosei
Posts: 1786
Joined: Wed Dec 14, 2016 3:59 am
GD Posts: 0
Has thanked: 185 times
Been thanked: 495 times

Re: Introducing SensAI Go

Post by jlt »

Wow these comparisons with cannibal infanticidal nurses are scary.
Independently of cheating, I'm wondering if intuition about who is ahead can be trained by immediate feedback as suggested by the OP. A way to check this would be to create an application which shows a position from a random game where one side is clearly ahead according to AI (say at least 75%) and ask the user to guess who is ahead. If the user's success rate improves with time, then it shows their intuition improved.
Post Reply