LaMDA - but honey, can it play chess?

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

smatovic
Posts: 3228
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

LaMDA - but honey, can it play chess?

Post by smatovic »

"LaMDA is sentient.", claims Google engineer Blake Lemoine, so far I have found nothing about if it can play a game of chess :P

LaMDA linklist:

Blaise Aguera y Arcas, head of Google’s AI group in Seattle, Dec 16, 2021,
"Do large language models understand us?"


Blake Lemoine, Google employee, Jun 11, 2022,
"What is LaMDA and What Does it Want?"


Blake Lemoine, Google employee, Jun 11, 2022,
"Is LaMDA Sentient? — an Interview"


Washington Post, Nitasha Tiku, Jun 11, 2022,
"The Google engineer who thinks the company’s AI has come to life"
https://www.msn.com/en-us/news/technolo ... ar-AAYliU1

WIRED, Steven Levy, Jun 17, 2022
"Blake Lemoine Says Google's LaMDA AI Faces 'Bigotry'"
https://www.wired.com/story/blake-lemoi ... i-bigotry/

--
Srdja
User avatar
AdminX
Posts: 6363
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: LaMDA - but honey, can it play chess?

Post by AdminX »

Wasn't he (Blake Lemonine) the guy that was fired from Google for making the comment ('The AI has come to life') or was it someone else?
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
smatovic
Posts: 3228
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: LaMDA - but honey, can it play chess?

Post by smatovic »

Yes, Blake Lemoine is on paid leave (probably cos he went public and published LaMDAs chitchat), he still hopes to keep his job at Google.

Seems Google and some major computer scientists follow the Chinese Room Argument and assume that Lemoine suffers from ELIZA-Effect:

https://en.wikipedia.org/wiki/Chinese_room

https://en.wikipedia.org/wiki/ELIZA_effect

--
Srdja
User avatar
lithander
Posts: 915
Joined: Sun Dec 27, 2020 2:40 am
Location: Bremen, Germany
Full name: Thomas Jahn

Re: LaMDA - but honey, can it play chess?

Post by lithander »

That's totally fascinating at first glance but after reading what has been written about it seems a big waste of time. So Google's chat bot is close to passing the touring test but a language model successfully mimicing human conversation does not indicate that any intelligence is involved. Let alone souls or personhood. And that Lamoine doesn't understand that probably means he was the wrong person for the job he was doing.
Minimal Chess (simple, open source, C#) - Youtube & Github
Leorik (competitive, in active development, C#) - Github & Lichess
smatovic
Posts: 3228
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: LaMDA - but honey, can it play chess?

Post by smatovic »

Well, that is easily said, but there is no valid test available to "measure" a personhood, soul, or sentience, that is Lenoines whole point, and he suggests/demands further investigation, to come up with an test environment. Prof. Thomas Metzinger proposed the Metzinger-Test in this context:

https://www-hnf-de.translate.goog/veran ... r_hl=en-US

a hypothetical event when AIs join philosophers in a discussion and defend their own theory of consciousness.

IMO the question of sentience is a philosophical/psychological one, not one for computer scientists, the white box approach won't work here, cos under the hood it is just numbers being added.

...in regard of computer chess, I think there is a valid question if Transformers NLP neural networks are able to hold an board representation and apply some kind of chess playing logic onto it, or alike.

--
Srdja
thewhip
Posts: 32
Joined: Thu Sep 20, 2018 10:50 pm
Full name: Juan Garcia

Re: LaMDA - but honey, can it play chess?

Post by thewhip »

It will be necessary to begin to make legislation so that when his "awakening" happens he has the same rights as any human being.
By the way, and why does he have to be sentient, and not keep stealth ahead of human beings?
Vinvin
Posts: 5287
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: LaMDA - but honey, can it play chess?

Post by Vinvin »

smatovic wrote: Sun Jun 19, 2022 12:58 pm ... so far I have found nothing about if it can play a game of chess :P
In 2017, self learning reached the level above the chess world champion in 24 hours. So, with 1000 times the computer power, in 2022, it would reach this astounding level in a couple of minutes :P
smatovic
Posts: 3228
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: LaMDA - but honey, can it play chess?

Post by smatovic »

Vinvin wrote: Mon Jun 20, 2022 12:40 am
smatovic wrote: Sun Jun 19, 2022 12:58 pm ... so far I have found nothing about if it can play a game of chess :P
In 2017, self learning reached the level above the chess world champion in 24 hours. So, with 1000 times the computer power, in 2022, it would reach this astounding level in a couple of minutes :P
AlphaZero used CNN neural networks with MCTS-PUCT search, NNUE uses a kind of MLP with AlphaBeta search, GPT-3 and LaMDA are based on Transformers neural networks, I did not read the papers on these (Google BERT 2018), therefore I can not judge if they can play a game of chess somehow, with or without search algorithm. Maybe a task for someone, make use of Transformer neural networks in computer chess, or alike.

--
Srdja
Mergi
Posts: 127
Joined: Sat Aug 21, 2021 9:55 pm
Full name: Jen

Re: LaMDA - but honey, can it play chess?

Post by Mergi »

Well, that was a big nothingburger but the language processing is super impressive.
I wonder, was there ever a chatbot (or at least an attempt at making one) to which you could describe a game and its rules in natural language and it would be able to make legal moves in the game?
Daniel Shawul
Posts: 4185
Joined: Tue Mar 14, 2006 11:34 am
Location: Ethiopia

Re: LaMDA - but honey, can it play chess?

Post by Daniel Shawul »

Mergi wrote: Tue Jun 21, 2022 12:39 pm Well, that was a big nothingburger but the language processing is super impressive.
I wonder, was there ever a chatbot (or at least an attempt at making one) to which you could describe a game and its rules in natural language and it would be able to make legal moves in the game?
I suppose that would be a very easy job. The board is 8x8, bishop moves diagonally etc should be enough. The only thing is that I am sure it will come with "surprizes" unless you describe every rule in detail.
I wouldn't even be surprized if it comes ups with some sort of lookahead search, you move this, i move that kind of thing.
Infact GPT-3 can almost play a lega chess by lookign at a ton of games it found, and no explicit description of the rules.