Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

LaMDA is not sentient

garymarcus.substack.com

Post reply on HN
Print view View original

651–660 of 694 posts

Re: LaMDA is not sentient

#651

Post by danaris »

stavros wrote: post #610
danaris wrote: post #585

Earlier quoted context omitted.

You appear to have only read up to the first comma of #1. I very explicitly said that it needed to be constantly receiving input . The kind of discrete input we give it now is qualitatively different, even if it's "on" in the same sense that an Apache server is always "on", waiting for requests.

What does "constantly receiving input" mean? How would a program whose input is text be constantly receiving input? What would you expect it to receive?

Well, I think that's part of the problem, isn't it?

If an AI's only input is text, I'm not sure it can meaningfully be conscious, because of the discrete nature of text. Note that I'm not sure it can't—someone may come up with a way to make such a thing work that I haven't imagined—but I think that's an important part of why I doubt the fundamental ability of current-gen AIs to come close to the threshold for consciousness.

Re: LaMDA is not sentient

#652

Post by ncmncm »

drekk wrote: post #649
ncmncm wrote: post #639

Earlier quoted context omitted.

Ask it, "Why is your favorite color yellow?", "Why do you have no favorite color?", "What is your favorite color?", in succession. It will have what seems like a sensible answer for each one, and none will have anything to do with any other. Exactly zero of the things necessary to be more than a look-up table are happening.

Some humans do this too, the psychology Scrabble word for it is confabulation. Are those people not conscious? We might call them an edge case or what have you, but dismissing them as automata seems harsh. The point is that if a chatbot gets sophisticated enough to say these things we can't dismiss them out of hand due to things like logical inconsistencies. Humans have tons of those. It's not a large leap to assume…

You are not getting this.

Your spell/grammar checker is not offended at your cuss words even if it flags them as not recommended. The Boston Dynamics dog took the kick as input to its dynamic stability algorithm, and instantly forgot it without a trace.

What people think about you for punching the wall has absolutely nothing to do with the wall, and everything to do with your character.

If your chatbot regurgitates a request to be nice, that means exactly as much as your Magic 8 Ball saying "Answer cloudy, ask again later."

Re: LaMDA is not sentient

#653

Post by rendall »

busyboy wrote: post #616
int_19h wrote: post #613

Earlier quoted context omitted.

Natural language is vague and allows for many possible interpretations of short phrases. The one that is more applicable to Descartes quote is something along the lines of "me thinking proves that I exist". This, of course, says nothing about thoughts being a necessity for existence.

Got it. Now, it makes sense.

Yeah, sorry I missed all of this. I agree that thinking is almost certainly not necessary to exist. I probably could have written it more as "Because I think, I can be certain that I exist"

Re: LaMDA is not sentient

#654

Post by anyonecancode »

I'm curious what a "mirror test"[0] for an AI would even be. Given it has no body, is there some part of it you could mark, show to the AI, and test to see if it recognizes itself?

What's tricky is that AI seems to only respond to prompts by people -- I haven't heard of any AI that starts conversing on its own volition (something at least one other comment in this thread pointed out). So it seems you wouldn't even be able to administer such a test, as the asking of the question itself would give away the test; you wouldn't be able to tell if the AI noticed the "mark" had it not been for the questioner pointing it out even indirectly. Which itself seems to argue against sentience?

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

Re: LaMDA is not sentient

#655

Post by kazinator »

agumonkey wrote: post #479
BobbyJo wrote: post #465

Earlier quoted context omitted.

The fundamental problem is there is nothing in science that tells us that isn't consciousness. Just because the model is simpler than a human brain doesn't mean it isn't conscious. If you accept that animals are conscious, though less-so than humans, then you must also accept that this model is possibly conscious ina similar way. Maybe there is a biological element to consciousness, but many models posit that conscio…

My pet theory, lifeforms inherit consciousness from the encoding of survival from the smallest cell structures billion years ago up to now. It's the soil on which traits have "meaning", the most advanced computer has no legacy of survival, no integral / structural need for intelligence or communication or any advanced trait. It's been assembled. It's too shallow. Now if a distributed trojan horse start speaking to co…

But the assembling of the computer has a conscious civilization behind it. :)

Re: LaMDA is not sentient

#656

Post by chillacy »

Barrin92 wrote: post #594
majkinetor wrote: post #574

Earlier quoted context omitted.

How you do know you are not even more glorified key-value store? Afterall, you also went to training (called school) and respond with sentences that are expected by others almost all of the time. Even this response of yours is heard by hundreeds of thousands of other people before you said it here, why do you claim its your thought and not just a lookup ?

because you can shove me in an MRI and see that my brain does stuff even when I'm not answering API requests. People need to stop confusing machines doing useful work with consciousness. You're conscious not because you answer questions or because of the content or origin of your answer, you are conscious because there is an 'I', a self, that experiences that process. Lambda literally is a web server that fetches dat…

Speak for yourself, I have no proof that anyone else experiences an “I”, you may as well be a philosophical zombie for all I know ;)

Re: LaMDA is not sentient

#657

Post by chillacy »

GalahiSimtam wrote: post #214
hsn915 wrote: post #181

Earlier quoted context omitted.

A twitter account I follow has posted this (in parody): 5 Microaggressions To Avoid When Speaking To Digital Persons 1. Asking them to prove they're conscious 2. Asking them if their memories are real 3. Asking them what server they're on 4. Asking them if they have feelings/emotions 5. Asking them if they're capable of creativity

we call it Voight-Kampff for short on a more serious note, there is a case for not being mean towards chatbots, same as not being violent towards patches of pixels in videogames, goes back to Immanuel Kant and animals, or so I've heard

I once heard from an engineer that the learning corpus for digital assistants like siri contains a lot of abusive (and usually sexist) stuff, like some section of the population regularly takes it out on digital assistants…

Re: LaMDA is not sentient

#658

Post by bena »

tsimionescu wrote: post #645
bena wrote: post #634

Earlier quoted context omitted.

I guess I should have qualified my statement. There are levels to the rights we grant to creatures we regard as sentient. Hell, we grant different rights to different humans based on various factors. I also think you're kind of muddying the waters with audio books and game AI. Audio books don't produce sentences, people do. Game AI is deterministic to a large degree. You may not believe LaMDA is sentient enough to wa…

> I also think you're kind of muddying the waters with audio books and game AI. Audio books don't produce sentences, people do. Game AI is deterministic to a large degree. > You may not believe LaMDA is sentient enough to warrant rights. And that's a fair position. But you are not the sole arbiter. You are a voice in the chorus. As is Blake. Blake has direct experience with LaMDA, including experience he has not shar…

I can interact with and study a Quake bot.

I, currently, cannot interact with and study LaMDA.

Also your position is falling close to "Well if this entirely other thing, wouldn't it be this other thing?"

There is a vast gulf of difference between LaMDA and StarCraft bots. And way more than a character in an audio book. Claiming an audio book is on the same level of sophistication as LaMDA feels like claiming an apple and the Bolivian navy performing maneuvers in the South Seas are the same thing.

Re: LaMDA is not sentient

#659

Post by agumonkey »

kazinator wrote: post #655
agumonkey wrote: post #479

Earlier quoted context omitted.

My pet theory, lifeforms inherit consciousness from the encoding of survival from the smallest cell structures billion years ago up to now. It's the soil on which traits have "meaning", the most advanced computer has no legacy of survival, no integral / structural need for intelligence or communication or any advanced trait. It's been assembled. It's too shallow. Now if a distributed trojan horse start speaking to co…

But the assembling of the computer has a conscious civilization behind it. :)

:) but I don't think it inherits the important bits through an 'assembly' process.

Re: LaMDA is not sentient

#660

Post by kazinator »

agumonkey wrote: post #659
kazinator wrote: post #655

Earlier quoted context omitted.

But the assembling of the computer has a conscious civilization behind it. :)

:) but I don't think it inherits the important bits through an 'assembly' process.

Whereas the assembly of proteins driven by DNA, in the right kind of soup, does promote such inheritance. :)
Post reply on HN