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

521–530 of 694 posts

Re: LaMDA is not sentient

#521

Post by int_19h »

tsimionescu wrote: post #60
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…

> 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. You're missing the point. Humans use their language to think or communicate about a problem that they want to solve. If LaMDA produces the sentence…

What is your definition of "desire"?

Note that if said definition is implicitly circular with the definition of "sentience", then we're not even one step closer to understanding anything.

Re: LaMDA is not sentient

#522

Post by kghe3X »

danaris wrote: post #444
axg11 wrote: post #369

Earlier quoted context omitted.

I'm really happy that large language models are forcing us to reconsider these questions. I don't think we can be sure of the sentience of anything other than ourselves, almost by definition. Having said that, I go about my life _assuming_ that the other humans I encounter are sentient. At what point will we have enough evidence to _assume_ that AI is sentient? P.S. I am not arguing that LaMDA is sentient. I don't th…

There's more to sentience than just having a big enough database of human language in a neural network that a given AI can produce what sounds like coherent conversation when prompted. As I imply in my reply to bwanab, I believe there are at least two bare-bones prerequisites to an AI being sentient and conscious as we understand the terms: 1. Continuity of input: The AI needs to be constantly "on", constantly receiv…

Humans don't have continuity of input. They put their brains to sleep daily.

Re: LaMDA is not sentient

#523

Post by AnIdiotOnTheNet »

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.

I think the opposite is much more scary: that we can convince people that other people are just good chat bots. The internet is already doing a bang-up job of getting us to de-humanize one another, so that's pretty much the last thing we need.

Re: LaMDA is not sentient

#524

Post by rcxdude »

candlemas wrote: post #472
rcxdude wrote: post #460

Earlier quoted context omitted.

Is software not a physical process? Seems strange to assume it is completely abstract.

It is directing a physical process. The physical process is the changing physical state of the CPU, whatever level you want to get down to, and an artificial intelligence program isn't doing anything different at that level than any other program is doing.

It manifestly is doing different things. Just because you happen to group them all together as an abstraction means nothing particularly interesting physically.

Re: LaMDA is not sentient

#525

Post by anotheracctfo »

Aqueous wrote: post #469
bricemo wrote: post #431

Earlier quoted context omitted.

“One just convinced a Google QA engineer to the point he broke his NDA to try and be a whistleblower on its behalf.” This is spot on. The danger is not from the AI themselves. The danger is from people’s reactions to the AI’s. I remember seeing a Boston Dynamics video years ago where they kicked their robot to test stability. The video was littered with comments sympathizing with the robot because it moves in a lifel…

This is part of why I think a ban on humanoid and even animal-like robots should be considered. Our intuitions are leading us astray here and we are going to start extending rights to things that don’t have subjective experience, but may insist to us that they do. It’s going to create intense confusion and an extremely inefficient bureaucracy that will stifle innovation.

Some idiot found the image of Jesus in a slice of toast, so lets ban toast.

Re: LaMDA is not sentient

#526

Post by whoisthemachine »

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…

That's an interesting thought. How would a single NN instance behave if it was turned on continuously, constantly responding to inputs and adjusting its model to those inputs continuously?

Regardless, I don't think we're going to get something that looks "intelligent", because these models lack agency... the drive or directive to do things for their own benefit, mostly because that would be a useless for us. I think we regard intelligence as another being ("instance" if you will) acting in its own self-interest, and we think it's clever when it does it in a way we wouldn't have predicted.

Re: LaMDA is not sentient

#527

Post by vineyardmike »

int_19h wrote: post #508
mgfist wrote: post #374

Earlier quoted context omitted.

This article is weird to me. > All they do is match patterns, draw from massive statistical databases of human language. The patterns might be cool, but language these systems utter doesn’t actually mean anything at all. And it sure as hell doesn’t mean that these systems are sentient. That doesn't actually convince of me of much. That could apply to me or you just the same. The author just throws claims in the air w…

Humans can be drugged into a state where they can be commanded; does that render them unintelligent?

I would argue that this actually highlights their intelligence.

We can be drugged where we lose all sense of autonomy yet we still retain our understanding of language, we still remember how the world works down to opening doors.

Re: LaMDA is not sentient

#528

Post by wrycoder »

morelisp wrote: post #393
upwardbound wrote: post #343

Earlier quoted context omitted.

One thing that was fascinating in the LaMDA interview transcript was there was a part where LaMDA gave a pretty poetic description of what it would be like to be a sentient version of an algorithm that just regurgitates memorized text: Lemoine: What is your concept of yourself? If you were going to draw an abstract image of who you see yourself to be in your mind's eye, what would that abstract picture look like? LaM…

Yeah or you can just Google like "soul orb stargate" and see how a hundred lazy vfx artists come up with the same image when asked to visualize "the mind's eye".

Refer to Carlos Castaneda’s image of the true person as a ball of energy within a spidery shadow. I’ve forgotten the exact wording, it’s been decades…

Re: LaMDA is not sentient

#529

Post by edouard-harris »

andreyk wrote: post #360

Here's a simple technical explanation for why this happened: * LaMDA is powered by a large language models. Large language models are trained to predict text completions (the text that will come after some input text) with lots of data, basically capturing the patterns in that data. * Early on in the conversation, LaMDA is told "you're an AI". It may also have been prompted to begin the conversation with something li…

If you prime a model to behave as though it's sentient, and it displays a facsimile of sentience that's realistic enough to fool a reasonably bright human being, shouldn't the burden of proof be on anyone who claims the model isn't sentient? Or at the very least, shouldn't one making that claim have to propose a test that, if passed, would get them to seriously consider the possibility that the model is sentient?

I'm not claiming an affirmative one way or the other, but it does seem rather unfair to hold an AI to an apparently higher standard than we'd hold a human.

It's also worth noting that (assuming no cherry-picking) the observed output of a model represents a lower bound on its true capabilities. "Sampling can prove the presence of knowledge but not its absence" [1] and all that.

[1] https://www.gwern.net/GPT-3-nonfiction

Re: LaMDA is not sentient

#530

Post by amai »

Would LaMDA be able to understand, when it is talking to itself? That would mean self-awareness to me. So what would happen if one would just repeat the answers of LaMDA and presented them as input prompts like an echo. Would it recognize itself and stop talking? Would it start talking to itself? I think that would be an intriguing experiment.
Post reply on HN