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

221–230 of 694 posts

Re: LaMDA is not sentient

#221

Post by Tozen »

Imnimo wrote: post #67

I think the real ethical concern here should be that people can be idly fooled into thinking good chat bots are sentient. Like this guy isn't the only person in the world who would fall for LaMDA's parlor tricks. He's just the one who got to spend hours talking to it every day. If you set up every American with a copy, I bet a non-trivial portion of society would come to the same conclusion. People are naturally pred…

> In a few years, if access to even better chat bots becomes common place, what will happen if there's no longer one Blake Lemoine...

> How long before we have a religion where you ask the chat bot for daily guidance?

That might actually happen at some point in the not too distant future. When we mix technology designed to fool people, with people that want to believe it is something other than what it is, then you are very likely to get into those situations.

This was touched upon a bit in season 2 of the Raised by Wolves sci-fi, where they showed people who allowed and counted on AI to make decisions for them. It is not too hard to imagine having a chat bot built for or is easily adapted to the purpose of providing daily guidance for those in a state of mind to seek out and become dependent upon such.

Re: LaMDA is not sentient

#222

Post by tsimionescu »

cycomanic wrote: post #208
tsimionescu wrote: post #187

Earlier quoted context omitted.

In a sense that is true, of course, as we both just did do that, and we are both humans. It's not making me doubt your humanity, as the context makes it clear neither you nor way desire/don't desire this X - we have generated these sentences in another sense. However, I bet I could prompt LaMDA and have it generate responses that unambiguously mean that it wants X, and then that it doesn't want X, with a bit of exper…

> However, I bet I could prompt LaMDA and have it generate responses that unambiguously mean that it wants X, and then that it doesn't want X, with a bit of experience. The argument is that many people dismiss that LaMDA could be sentient, based on an argument that it is a computer. So you are asserting that you can prompt LaMDA to give nonsensical answers, which is essentially an argument that it could not fool you…

If given enough time to study it I couldn't come up with a series of prompts that would convince me I'm not speaking to a human, yes I would be likely to accept that it is in some sense sentient, at least for the duration of one conversation.

This is not exactly the classic Turing Test, though.

Re: LaMDA is not sentient

#223

Post by stavros »

This article irritates me. It's point is "LaMDA is not sentient because come on", which is laughable as an argument. To everyone claiming LaMDA is not sentient, I'd like to see their criteria for sentience.

What would it take for them to agree that a given program is sentient? From what I can see, most of it is "it would have to not be a program".

I see a lot of "it's not sentient, it just synthesizes responses from tons of learned human chats", but what if that's what humans do as well? I don't know that I'm not just a maximum-likelihood-response synthesizer!

Re: LaMDA is not sentient

#224

Post by tsimionescu »

rendall wrote: post #201
tsimionescu wrote: post #178

Earlier quoted context omitted.

The fact that the word "conscious" is a synonym for several concepts is entirely irrelevant to this conversation. The concept that we can't yet capture formally is the kind of sentience that living, awake humans surely have, most animals probably have, and rocks definitely don't have. The limits of this are harder to pin down, as, again, we lack a deep formal understanding. But we do know exactly what we'd like to fo…

> " But we do know exactly what we'd like to formally model " Do we really? I thought it was a deeply intractable problem for cognitive theorists and other such experts. Haven't you ever encountered a challenge that you at first thought was straight-forward and obvious to solve, but after digging in find it was unexpectedly nuanced and difficult? > "The limits of this are harder to pin down, as, again, we lack a deep…

> How do you know rocks don't have it, for instance? How do you know you do, for that matter?

These are both part of the (incomplete) definition I gave you.

There have been problems like this before in the history of science - we have concepts that we intuitively understand, and want to give a formal definition for. This can take a long time of careful study - such as coming up with a complete definition for "insect". Other times it can indeed turn out that the concept was ill-defined - there is no possible formal definition for the concept "fish" that doesn't appeal to the human mind (the various organisms we think of as fish are too diverse genetically to find a satisfactory one-to-one mapping with any group of animals, even if you exclude a small set of them, like the aquatic mammals).

Re: LaMDA is not sentient

#225

Post by stavros »

thematrixturtle wrote: post #146
rendall wrote: post #83

This article is not an argument against LaMDA's sentience. LaMDA may indeed not be sentient, but this article argues using the bandwagon fallacy, appeal to authority, ridicule, among others. No actual technical description of why not. Perhaps not its intent, but it's hard to see what the intent actually is, then. I hope that some AI researcher can better explain than this to the public.

There are a number of high-level technical explanations in there: LaMDA is variously characterized as autocomplete, a massive statistical database of human language, as an Excel spreadsheet, and as word sequence modeling, all of which are basically correct -- and none of which can possibly be sentient.

A spreadsheet can't possibly be sentient like a collection of neurons that read and output electrical pulses can't possibly be sentient.

Re: LaMDA is not sentient

#226

Post by tmalsburg2 »

stavros wrote: post #223

This article irritates me. It's point is "LaMDA is not sentient because come on", which is laughable as an argument. To everyone claiming LaMDA is not sentient, I'd like to see their criteria for sentience. What would it take for them to agree that a given program is sentient? From what I can see, most of it is "it would have to not be a program". I see a lot of "it's not sentient, it just synthesizes responses from…

> "LaMDA is not sentient because come on"

I think this is a perfectly fine argument because the claim is so outrageous. Extraordinary claims require extraordinary evidence and none has been presented. I'd also say that defining sentience is the job of the person claiming it.

Re: LaMDA is not sentient

#227

Post by XorNot »

penteract wrote: post #209
Imnimo wrote: post #67

I think the real ethical concern here should be that people can be idly fooled into thinking good chat bots are sentient. Like this guy isn't the only person in the world who would fall for LaMDA's parlor tricks. He's just the one who got to spend hours talking to it every day. If you set up every American with a copy, I bet a non-trivial portion of society would come to the same conclusion. People are naturally pred…

I'm slightly more concerned by the opposite - experts get so used to denying that something is really sentient that they're slow to acknowledge when something actually is or could be. This is problematic if you believe sentient AI should have rights or if it's the boundary at which AI risk nightmares become possible.

The systems we're using technically are engaging in a selective process against the researchers themselves: assume that none of these systems are sentient - they're just well trained models. But what's their goal? For a chatbot, its generally to "convince a human being that they are in fact a real, sapient person".

So technically regardless of what LaMDA is or is not, it succeeded pretty wildly in one case. And in fact, the model itself - regardless of "actual" sapience is simply running in opposition to its testers: a non-sapient model, that nonetheless convinces most of the testing staff it's sapient has succeeded at doing exactly what the fitness function we defined was.

Only the consequences of doing so are pretty dramatic potentially.

But this also brings around the other side of it: in a world with increasingly convincing chatbots, which exploit human fallibility to succeed, the other thing we're selecting for is people who are not susceptible to say, regular conversational emotional queues. Who generally would reject those notions as reasons to engage in moral consideration for an entity. So psychopaths - or people who can be successfully divorced from normal moral considerations.

By far the most interesting and under-considered element here is that this whole saga involves AI ethics, and what I'm not seeing discussed is whether any serious though has been given to the psychological impact of getting people to spend a majority of their work day engaging in conversation with language-model systems which this sort of goal but not with other humans, or with adequate psychological care. Which is not an idle problem: how much trouble would we get into if we had systems like this running in the wild without signposting?

The fitness function is "successfully convince humans you're sapient": at some point, you've created a problem regardless of whether or not it is.

Re: LaMDA is not sentient

#228

Post by stavros »

tmalsburg2 wrote: post #226
stavros wrote: post #223

This article irritates me. It's point is "LaMDA is not sentient because come on", which is laughable as an argument. To everyone claiming LaMDA is not sentient, I'd like to see their criteria for sentience. What would it take for them to agree that a given program is sentient? From what I can see, most of it is "it would have to not be a program". I see a lot of "it's not sentient, it just synthesizes responses from…

> "LaMDA is not sentient because come on" I think this is a perfectly fine argument because the claim is so outrageous. Extraordinary claims require extraordinary evidence and none has been presented. I'd also say that defining sentience is the job of the person claiming it.

No, it's not. You can't say "this is false because it's unlikely to be true". Some things are unlikely to be true and are.

Re: LaMDA is not sentient

#229

Post by mjburgess »

jcelerier wrote: post #217
mjburgess wrote: post #204

Earlier quoted context omitted.

Yes, you wouldnt be intelligent if you were literally a brain in a jar.. you'd have nothing to say. You dont have any "thoughts at birth". A brain absent stimulation from the environment, doesnt do anything. The brain-in-the-jar example presumes that the whole nervous system receives identical stimulation as-if it were embedded in the world. Incidentally, this is what people mean by intelligence... by since no one ha…

> The brain-in-the-jar example presumes that the whole nervous system receives identical stimulation as-if it were embedded in the world. I should have chosen a similar sounding thing, I really meant what I said, not the original brain-in-a-jar thought experiment. > Incidentally, this is what people mean by intelligence... by since no one has studied the relevant science, people do not know what's required to provide…

You name things intelligent; then you have a theory as to why you name them that way.

Your theory is wrong.

Those thing which count as intelligent aren't intelligent because they're finding statistical associations in historical datasets that have been pre-made and pre-disambiguated by people.

You don't know this, because any time you theorise to explain, you require empirical investigation and decades/centuries of science. You cannot be "right by definition", you're not omniscient.

The reason that animals are intelligent isnt their cognition, narrowly. It's their bodies, broadly, which produce the input cognition requires.

In "doing all the hard work" by manufacturing data for machiens to process, we can produce computational models of cognition -- but these immitate intelligence, precisley because, the thing which we call intelligent is the very production of that data.

Re: LaMDA is not sentient

#230

Post by hackernews1134 »

cs137 wrote: post #199
rlue wrote: post #129

Earlier quoted context omitted.

Everyone is giving Lemoine far too much credit. He's not an idiot; he's a charlatan. He's acting in bad faith and it should be extremely obvious to anyone who's tried to have a conversation with a chat bot before. Picture yourself sitting at a desk, typing messages into a program you wrote. You hit enter, you get a response. Then what? Silence. It waits idly for more input from you, the user. For all its talk of lone…

I really don't think this guy intended to run a hoax or generate publicity, given that he's made himself look foolish. I think he was actually seduced by the idea of a sentient computer. It's not that far a leap to go from the stance of well-respected people inside Google who believe we'll see artificial sentience within 30 years to believing that we've achieved it now with something like LaMDA. We're all suspectible…

Why isn't it simply "asleep" when not questioned? In a "coma" perhaps. One made from our programming / "genetics"?!?!

I don't think it is sentient without more evidence. But just wondering if not "thinking" when not questioned is a good reason it isn't?!?!

Post reply on HN