Alan Turing was a visionary.
-
@tiotasram @raymaccarthy @futurebird huh. very interesting nuance, thanks for that.
-
All of this handwringing about conciseness is ultimately about morality. Should you feel bad about crushing a bug? How bad should you feel?
Destroying beautiful things, destroying complex things, especially complex things that you don't understand strikes me as significant.
It's why you feel something when you see a mandala erased from the sand. It's why that erasure is incorporated into the tradition.
Sweeping the floor is not the same if there is a mandala.
@futurebird @ireneista so, to be entirely honest here, I don't think Alan Turing's "Imitation Game" (the original name for the Turing Test) was meant to determine consciousness. The Imitation Game was his way of answering the question "Can machines think?", which I feel like is a very different question, especially in 1950.
I feel like it would be appropriate to say that many computers of our modern day do something you could call "thinking", even if they aren't really an AI system (take any programmed application you use to perform difficult automated tasks with. Perhaps Excel is a good example).
I recently read his paper where he introduced the concept, and it was incredibly succinct, and to me had a lot more to do with *computers* than it did with *AI* (though it of course dabbled in both). I think he was trying to demonstrate the potential of computers to an audience who really had only ever seen them as clunky, single purpose calculators that lacked elegance.
Also fun fact: Turing speculated that by the year 2000, we ought to be able to produce a machine which has 1 whole entire Gigabyte of storage, and using that, we could get it to play the Imitation Game sufficiently. Now we've got chat models that suck at thinking, and take 100+ gigabytes to do it....
-
@raymaccarthy @futurebird but yeah, we played with Eliza as kids, learned its ins and outs, read a bit about the history, and kind of thought everyone had learned the lesson to not take the machine too seriously just because it's generating English text...
........ apparently not everyone paid attention though
@ireneista @raymaccarthy @futurebird Eliza and LLM models based on transformers are not at all the same. One of the first programs I typed in in 1980 was Eliza. Keyword matching and canned response is not prediction.
-
Alan Turing was a visionary. Super-perceptive computer scientist and it annoys me to no end that what he's most famous for outside of computer science is the "Turing Test."
He gave one of the first and most succinct accounts of how a computer should work and they still work that way to this very hour as I type.
Talk about Turing Machines more and Turing Tests less.
@futurebird Thing that annoys me is when people say passing the Turing test = consciousness. After he thought about it for 5 minutes he specified that all it meant was being able to fool a human, and that's not special. Patterns on a piece of toast can do that.
-
Alan Turing was a visionary. Super-perceptive computer scientist and it annoys me to no end that what he's most famous for outside of computer science is the "Turing Test."
He gave one of the first and most succinct accounts of how a computer should work and they still work that way to this very hour as I type.
Talk about Turing Machines more and Turing Tests less.
@futurebird and very readable papers to this day. I like his writing. Very grounded.
-
@ColinHaynes @futurebird Unfortunately that's a terrible film in terms of historical accuracy and representation of Turing's character.
I'd recommend Breaking the Code if you want visual media which better encapsulates Turing as a person.
-
@ireneista @raymaccarthy @futurebird Eliza and LLM models based on transformers are not at all the same. One of the first programs I typed in in 1980 was Eliza. Keyword matching and canned response is not prediction.
@noplasticshower @ireneista @futurebird
No, they don't work the same. However that doesn't matter. I suggested the biggest limitation of the originals was the built in data, The current ones are still amusing toys and it's a scam on investors and users to claim they are actually useful. It's hype and self-delusion. -
The problem with developing a "test for conciseness" is we do not have a definition for what it is that would allow such a test to work with other people who we can presume to be conscious (if conciseness can be well defined)
I think we should retreat to simpler questions. Here is one:
Is it possible for pain and suffering to exist without conciseness?
@futurebird @ireneista oh yes. Pain is just a damage signal tied to specific, rather urgent "get away from the damage" incentives. Consciousness, as far as I can tell from my kindergarten level understanding, is probably a sort of mental reflection function?
-
@futurebird @ireneista oh yes. Pain is just a damage signal tied to specific, rather urgent "get away from the damage" incentives. Consciousness, as far as I can tell from my kindergarten level understanding, is probably a sort of mental reflection function?
@SomeVeganCheeseIsOk @futurebird the question is complicated by the fact that many attempts to define "consciousness" describe things that humans don't even do
-
@SomeVeganCheeseIsOk @futurebird the question is complicated by the fact that many attempts to define "consciousness" describe things that humans don't even do
@SomeVeganCheeseIsOk @futurebird it seems like a topic that it should be possible to study seriously, and we've read research that is making serious efforts, but none of it has felt persuasive to us
-
@SomeVeganCheeseIsOk @futurebird it seems like a topic that it should be possible to study seriously, and we've read research that is making serious efforts, but none of it has felt persuasive to us
@SomeVeganCheeseIsOk @futurebird people just bring an awful lot of preconceptions about it, which makes it really hard to talk about
-
@SomeVeganCheeseIsOk @futurebird the question is complicated by the fact that many attempts to define "consciousness" describe things that humans don't even do
@ireneista @futurebird the reading I've done on consciousness was fascinating - the fact that it comes significantly *after* our reactions to things, and provides a thought train that justifies those reactions, was wild to learn.
-
@ireneista @futurebird the reading I've done on consciousness was fascinating - the fact that it comes significantly *after* our reactions to things, and provides a thought train that justifies those reactions, was wild to learn.
@SomeVeganCheeseIsOk @futurebird yes for sure!
-
@tiotasram @ireneista @futurebird
It's not really a test because it's absolutely subjective and there is no scoring criteria.
You know how many "romances" written for women by people with female pen names are actually so? Maybe 70%. The idea of convincingly playing a gender role is nothing to do with computer programs. It's a worthless thought experiment. Many are actually read by men too. Chicklit is a demeaning phrase.
People have done it perfectly, badly and deliberately as entertainment. -
@ireneista @futurebird
The Turing Test (not a real test) was never serious.
Alan Turing died in 1954. Chess, thought originally to need AI, didn't. He wrote one of the first.
The Eliza Chatbot was developed 1964 to 1967.
13 yrs?
The main limitation was that the data could not easily be extended. It "passed" the Touring test for some naïve users. The Doctor version is in Linux emacs. Run it, hit escape, type x and then type doctor.
The current LLMs have huge datasets, so seem more realistic.@raymaccarthy Side remark: Alan Turing killed himself because the laws against homosexuality were enforced against him. We should take the time to use that memory to keep fighting against fascist laws. @ireneista @futurebird
-
@SomeVeganCheeseIsOk @futurebird people just bring an awful lot of preconceptions about it, which makes it really hard to talk about
@ireneista @futurebird for brain funsies, I really liked "the power of habit" and "the man who mistook his wife for a hat"
-
@raymaccarthy Side remark: Alan Turing killed himself because the laws against homosexuality were enforced against him. We should take the time to use that memory to keep fighting against fascist laws. @ireneista @futurebird
@carl @raymaccarthy @futurebird oh MOST DEFINITELY
some of the younger queer people on here have come up with the slogan "make sure to be extremely gay on the computer or Alan Turing died for nothing".
it's not, like... we have some professional/activist experience in designing slogans and that's not one we'd have picked, there are many problems with it, but it sure does speak to an emotional truth.
-
@carl @raymaccarthy @futurebird oh MOST DEFINITELY
some of the younger queer people on here have come up with the slogan "make sure to be extremely gay on the computer or Alan Turing died for nothing".
it's not, like... we have some professional/activist experience in designing slogans and that's not one we'd have picked, there are many problems with it, but it sure does speak to an emotional truth.
@carl @raymaccarthy @futurebird it's history everyone should know. powerfully, powerfully relevant to today.
-
@raymaccarthy Side remark: Alan Turing killed himself because the laws against homosexuality were enforced against him. We should take the time to use that memory to keep fighting against fascist laws. @ireneista @futurebird
@carl @ireneista @futurebird
That's true, even if in fact he poisoned himself by accident. Look up electroplating with gold; it seems to have been a hobby. Cyanide based chemicals. -
@noplasticshower @ireneista @futurebird
No, they don't work the same. However that doesn't matter. I suggested the biggest limitation of the originals was the built in data, The current ones are still amusing toys and it's a scam on investors and users to claim they are actually useful. It's hype and self-delusion.@raymaccarthy @ireneista @futurebird I am sorry, but I disagree with your characterization. And yes, I work on this directly and with a great deal of scientific skepticism (see https://berryvilleiml.com/). I wrote my first neural network 9 years after Eliza in 1989 and trained it to beat along with music.
There are many reasons that LLMs are like models of alien intelligence because they are not like us. But they are more like us than Eliza with a huge database. Lol.
