DecodeChess has already done this.
I don't think ChatGPT is able to play chess yet, but I think Wikipedia is not a credible site since users can change data.
<Comment deleted by user>
@RyanLiu2020 said in #13:
> I think Wikipedia is not a credible site since users can change data.
Society began to collapse since petroglyphs stopped being fashionable.
> I think Wikipedia is not a credible site since users can change data.
Society began to collapse since petroglyphs stopped being fashionable.
data? Yes words can fail us.. but should we drop them altogether?
The success of ChatGPT shows that an AI chess coach would be very nice indeed. ChatGPT isn't very good at that, but maybe another tool could be, if it were developed.
@pawngrid said in #17:
> The success of ChatGPT shows that an AI chess coach would be very nice indeed. ChatGPT isn't very good at that, but maybe another tool could be, if it were developed.
your thesis is that AI will be able to help with chess coaching in the future? someone get this person a nobel prize!
> The success of ChatGPT shows that an AI chess coach would be very nice indeed. ChatGPT isn't very good at that, but maybe another tool could be, if it were developed.
your thesis is that AI will be able to help with chess coaching in the future? someone get this person a nobel prize!
Can ChatGPT find all PGN that contain certain annotation keywords, like "2 bishops" or a set of them.
It would not need to understand the positions out of those. This is not like a evolved google search is it. Or behind its chat abilities is indeed an evovled whole internet search or indexing or addressing innovation? Its babble must have some basis that way?
It would not need to understand the positions out of those. This is not like a evolved google search is it. Or behind its chat abilities is indeed an evovled whole internet search or indexing or addressing innovation? Its babble must have some basis that way?
I only read the last bit now about feeding a full information representation of position into some model part of the bigger model.
But how is the big data made of position-less PGN, where most annotations are likely to come from, going to become aware of that angle. What is the amount of PGN data with annotations that do not start from the standard initial position, or that for some reason do make the initial FEN explicit data.
I do not know about the specific internals for chat GPT. Assuming big data from internet strings, and humongous possible correlations within with deep enough already natural languaged NN architectures. I might even consider that like us, some meaning could emerge from work associations with some grammar providing some verbing to create some sense of causality within the possible strings building blocks (words, moves).
Putting the ability in the model, for meaning, may not be enough, if the data itself never touches it.
In most of our chess strings, the elephant in the room is the board, and the fun things we can move on it... not just being on some amorphous list of coordinates.. but a spatially ordered grid. Where the words to associate would be positions.
I am speaker without knowing the details, just trying to put my 2s together.
In general. Does Big data contain mechanistic** data?..
** in order to be concise and expressive I will stick to that formulation :). I do like that holy grail communication criterion.
But how is the big data made of position-less PGN, where most annotations are likely to come from, going to become aware of that angle. What is the amount of PGN data with annotations that do not start from the standard initial position, or that for some reason do make the initial FEN explicit data.
I do not know about the specific internals for chat GPT. Assuming big data from internet strings, and humongous possible correlations within with deep enough already natural languaged NN architectures. I might even consider that like us, some meaning could emerge from work associations with some grammar providing some verbing to create some sense of causality within the possible strings building blocks (words, moves).
Putting the ability in the model, for meaning, may not be enough, if the data itself never touches it.
In most of our chess strings, the elephant in the room is the board, and the fun things we can move on it... not just being on some amorphous list of coordinates.. but a spatially ordered grid. Where the words to associate would be positions.
I am speaker without knowing the details, just trying to put my 2s together.
In general. Does Big data contain mechanistic** data?..
** in order to be concise and expressive I will stick to that formulation :). I do like that holy grail communication criterion.