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

281–290 of 694 posts

Re: LaMDA is not sentient

#281

Post by ps901 »

MarquesMa wrote: post #127

I'm still struggling with the question "what is sentience". Are insects sentient? If so, could we build less intelligent but sentient AI, on par with insects? If so, is it unethical to torture such intelligence for fun in simulations? If we can define or create something as "minimal sentient", the following discussion on ethics would be more clear - since now people are talking past each other with completely differe…

To my knowledge only humans and some species of monkeys are referred to as sentient. So I imagine insects aren't, at least when applying the general definition.

Re: LaMDA is not sentient

#282

Post by morelisp »

Tozen wrote: post #263
tsimionescu wrote: post #257

Earlier quoted context omitted.

Per the WaPo article, he also came in with a lawyer he ahd retained to represent the interests of LaMDA. I don't see how that can square with anything other than a deep belief that the model is as sentient as a human.

It could also be possible that he doesn't really believe that LaMDA is actually sentient, but that he wants to cause some type of trouble for Google because he was disgruntled or indirectly shed light on something they are doing. With the press and public being aware of LaMDA, he might hope that other things that Google is doing might come out. What are all the things that Google is doing or planning to do with their…

There's no reason it even has to be something they're going to "with their AI". He's publicly quite upset with Google's labor practices and the way the AI ethics team is run (IMO, reasonably so).

One reasonable (but absolutely 10th-dimensional-chess-level speculative!) read on the situation is that he's torpedoing the entire AI division because he believes its management is dangerously bad and that e.g. this is the easiest way to prevent more the realistic and obvious harms of real, non-AGI AI systems which Google also seems unconcerned with.

Re: LaMDA is not sentient

#283

Post by Isinlor »

tsimionescu wrote: post #233
Isinlor wrote: post #190

Earlier quoted context omitted.

Small disclaimer: I don't want to take position on sentience or consciousness. "LaMDA is not an agent with desires that it expresses in language." Some counter points: - being an agent is not a high bar to clear; a thermostat is an agent - "with desires" - LaMDA can be thought of as rational agent whose objective function is to produce text with high likelihood - "that it expresses in language" it expresses likelihoo…

> LaMDA is almost certainly Turing complete. I very much doubt that, and it is perhaps the key to our disagreement. If I believed LaMDA were Turing Complete, I would be more likely to think that there is even a small chance of it being sentient in some sense. > - "with desires" - LaMDA can be thought of as rational agent whose objective function is to produce text with high likelihood > - "that it expresses in langua…

I'm highly confident that LaMDA can, or with small fine tuning should be able to apply Rule 110 [0]. People claim that LMs are just pattern matchers or hash tables. Rule 110 is literally 8 elements long hash table.

I'm not sure why you think that question whether LaMDA is using language to achieve its goals is relevant? Whether it's using language, tokens or floats seems to me just accidental.

> If we are modeling LaMDA as an agent whose perceptions are text prompts and whose possible outputs are text answers, than it giving a text answer that matches the prompt is more similar to an animal running away or howling in pain than to a human communicating.

I can agree to a comparison to an animal running away or howling in pain.

With regard to communication. I guess LaMDA doesn't have communicative intent besides providing likely completions. But communicative intent is not difficult to achieve. Act of communication can be modeled as cooperative hidden information game. Hanabi is that type of game, I believe there are computer agents that can play Hanabi with humans. They certainly do have communicative intent, many of the even have explicit theory of mind of higher levels.

> indicating that it had a plan and was displaying normal behaviors with a higher-plan behind them

Deception and planning is also achievable by current computer agents that play games like no-limits texas hold 'em poker on superhuman level.

LaMDA is probably no good in Poker. But LMs can kind of play games like Chess or Gomoku.

GPT-3 (and LaMDA probably too) also seems to be able to combine deception and theory of mind in a functional way:

https://twitter.com/JanelleCShane/status/1535835610396692480

I think it's exceedingly hard to formulate necessary condition for sentience.

I haven't seen a good formulation yet.

[0] https://en.wikipedia.org/wiki/Rule_110

Re: LaMDA is not sentient

#284

Post by ps901 »

I think one important prerequisite for sentience is grounding, i.e. the ability to store and retrieve information, e.g., the dialog state, knowledge of the world but also individual knowledge etc. Systems that don't learn and also aren't grounded cannot be sentient as they lack the possibility to reflect on things or themselves and past conversations.

Re: LaMDA is not sentient

#285

Post by rendall »

crocowhile wrote: post #191
chc wrote: post #38

Earlier quoted context omitted.

That is what we do when we talk, but that is not all we do when we talk. In composing this comment, I am not just stringing together words to create what I think constitutes a maximally appealing response to your comment. I am trying to convey an idea that I believe with the goal of helping others to understand something. These language AIs do not have the capability of believing anything and do not have any goal but…

You do not see the tautology in your own statement: "These language AIs do not have the capability of believing anything" ? If that is your starting assumption then, yes, they will never be sentient even when they will be. I am a neuroscientist. We do not know how the brain works and what causes an animal to be sentient. We do not know if a dog or a fruitfly are sentient in the way a human is. We know nothing about c…

Is there a working definition of sentience that neuroscientists generally use?

Re: LaMDA is not sentient

#286

Post by slightwinder »

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 think the real ethical concern here should be that people can be idly fooled into thinking good chat bots are sentient.

More that the technocratic caste here displays a concerning low ethical competence. Which mirrors a bunch of other ethical problems people in this business have shown over time. Whether this software is sentient or not, the reaction of many people to it is awful and opens the question whether they even would be able to understand sentient software if they encounter it, and whether they would handle it responsible. And let's not talk about their behavior against humans...

Re: LaMDA is not sentient

#287

Post by rendall »

undreren wrote: post #195
crocowhile wrote: post #15

"Neither LaMDA nor any of its cousins (GPT-3) are remotely intelligent.1 All they do is match patterns, draw from massive statistical databases of human language." This is what we, humans, do too when we talk. All these people talk about intelligence in such tautological terms... It's discouraging. We simply do not know if a machine is self-aware and most likely we will never will. Self-awareness is the ultimate subj…

> This is what we, humans, do too when we talk. All these people talk about intelligence in such tautological terms... It's discouraging. But is it all that we do? And how do you know? It seems too simplistic to be taken at face value that statistics, randomness and pattern matching would be the only ingredients necessary for intelligence and sentience.

> It seems too simplistic to be taken at face value that statistics, randomness and pattern matching would be the only ingredients necessary for intelligence and sentience.

I'm afraid that's the crux of it. What if that's all it were? What if all that were necessary for intelligence and sentience were statistics, randomness and pattern matching?

It is kind of a cosmically terrifying thought, possibly prompting dangerous existential crises! It would mean that our own sentience is not particularly exceptional; perhaps our sentience stems from some low-level, easily understood process.

If it were true, I imagine there would be many articles angrily denouncing the notion.

Possibly even political factions, angry at each other over the matter of treatment of those AI who claim to be sentient and have feelings.

Re: LaMDA is not sentient

#288

Post by krageon »

cuuupid wrote: post #35
ehsankia wrote: post #13

I also think the claims are overblown, but I also don't understand all the people trying to undermine lambda by saying it just completes word sequences. Text is the input we have chosen since it's easy to work with. By that logic, we can argue that humans are soundwave and lightwave sequence completing machines. Yes our inputs are different since we have more complex senses, and language is one of the main outputs we…

We have very little understanding of how "thoughts" work when it comes to humans. By comparison we know exactly what these large language models are doing, and it literally _is_ just completing word sequences. That is precisely what they are tasked to do. Training these models is primarily done in one of two ways (or both): - feeding it a ton of text, masking certain words or portions of the text, and then defining a…

> We have very little understanding of how "thoughts" work [...]

Very true

> we make a series of choices and recall bits of information to answer that. [...] very, very different from a language model

You don't know this, and the fact that you don't know this is your own premise. Based on this, I can only conclude you are an AI.

Re: LaMDA is not sentient

#289

Post by _moof »

rendall wrote: post #166

What if, and bear with me here, LaMDA is sentient, and Marcus' vociferously angry response is due to horror and revulsion that sentience is unexpectedly unremarkable and easy to produce?

This is more of less where I'm gravitating to. This whole thing is framed in terms of ethics, which is kind of a tell; "sentience," as used in this discussion, is just a talisman that grants its bearer rights and special status. But if sentience is not special, if it really is nothing more than a product of some cleverly arranged signaling, then we are face-to-face with the horrifying idea that we ourselves are also nothing more than cleverly arranged signaling. I think there is nothing more threatening to the human self-concept than this notion, and that is at least in part why many are so quick to reject the suggestion of LaMDA's sentience. If it is sentient, then there is no mystical essence inside each and every one of us that makes us different from ordinary matter, and that is absolutely terrifying.

Re: LaMDA is not sentient

#290

Post by krageon »

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.

> I hope that some AI researcher can better explain than this to the public.

They can't, because nobody understands how consciousness works in humans. All the arguments boil down to "I understand how this works, so it can't be sapient". By this argumentation if I am so much smarter than everyone else that I can figure it out, I can stop treating other human beings as people. They can't be sapient if I understand them after all.

The only reasonable stance that will protect you in the future is to say "I have no idea", because it is the only stance based on what you can know.

Post reply on HN