Jump to content

Deeply Confused Google Engineer comes to believe that Google's large language model is sentient, update: Deeply Confused Google Engineer has been put "on leave"

legend
 Share

Recommended Posts

Commissar SFLUFAN

Commissar SFLUFAN

Posted
Posted

For what it's worth, Stephen Pinker has weighed in:

 

Quote

Ball of confusion: One of Google's (former) ethics experts doesn't understand the difference between sentience (aka subjectivity, experience), intelligence, and self-knowledge. (No evidence that its large language models have any of them.)

 

 

From Gary Marcus:

 

https%3A%2F%2Fpbs.substack.com%2Fmedia%2

Nonsense on Stilts - by Gary Marcus

GARYMARCUS.SUBSTACK.COM
No, LaMDA is not sentient. Not even slightly.

 

Quote

 

Nonsense. 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. 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.

 

Which doesn’t mean that human beings can’t be taken in. In our book Rebooting AI, Ernie Davis and I called this human tendency to be suckered by The Gullibility Gap — a pernicious, modern version of pareidolia, the anthromorphic bias that allows humans to see Mother Theresa in an image of a cinnamon bun.

 

 

Link to comment
Share on other sites
legend

legend

Posted
  • Author
Posted
2 hours ago, TwinIon said:

I agree with @legend that there isn't any reason to suspect this thing is sentient, it did make me think that it would be pretty difficult to tell the difference between a system that actually is and one that is just the world's best chatbot. I honestly don't know how one might meaningfully make a distinction. I agree with @LazyPiranha that I'd expect an actual AGI to respond to things in a decidedly non-human way. So it would need to be human enough that we recognize its' intelligence, but at the same time you wouldn't expect them to always have a perfectly human response to every query.

 

I think it'll be fun to be able to talk to something like lambda. I wonder how resource intensive a conversation is right now and how far away they are from some kind of commercial application.

 

FWIW, I think evaluating "sentience," "consciousness," etc. after the fact is both a doomed and unnecessary process. We know quite well what cognitive faculties we've put into AI systems. There's no mystery that needs empirical evaluation after the fact. This is also why I, and so many other people in the field who are speaking out against this, can definitively say that Google's LLM is not sentient. We know, from first principles, what kind of cognitive faculties were built and what they don't do. It's important to note that despite what scifi stories often describe, AI isn't going to magically wake up one day. It's a scientific and engineering discipline. If we hope for magical moments, we'll never succeed. That's just not how science and engineering works. In fact magical moments is decidedly antithetical to science :p 

 

Philosophers will undoubtedly continue to debate what we even mean by "sentient" etc., and it's always possible to move the goal posts, but none of our definition changes will affect what it is that we actually built and our knowledge of that. Most people ponder whether our AI have sentience or consciousness purely out of being unsure what we mean by that to begin with rather than because of uncertainty of what the system does. There's also room of course for people to debate whether there's some metaphysical properties that AIs in conventional computers lack; whether they're philosophical zombies, but those are largely uninteresting questions from an AI perspective because they don't have anything to do with function of the system, which is what we care about. Even if I bought into the notion that philosophical zombies could exist (I don't), an AI would still be a success if we *only* built philosophical zombies that otherwise had all the cognitive faculties we want.

 

What does and will continue to require empirical evaluation is the capability of our systems on real world problems, because it's not at all clear how to judge difficulty of tasks with respect to the learning method a priori. For example, you need to test vision classification algorithms empirically, because we don't really know how to model the hardness of vision classification (which is in turn why it's a use case for ML at all!).

Link to comment
Share on other sites
Jason

Jason

Posted
Posted
5 hours ago, Commissar SFLUFAN said:
2637.jpg?width=1200&height=630&quality=8

Google engineer put on leave after saying AI chatbot has become sentient | Google | The Guardian

WWW.THEGUARDIAN.COM
Blake Lemoine says system has perception of, and ability to express thoughts and feelings equivalent to a human child

 

 

 

Silencing the guy who tries to warn everyone by dismissing him as a lunatic is how a whole hell of a lot of movies start...

Link to comment
Share on other sites
legend

legend

Posted
  • Author
Posted
22 minutes ago, Jason said:

 

Everyone thought Daniel Jackson was a nutjob. :nottalking:

 

 

Quote

“But the fact that some geniuses were laughed at does not imply that all who are laughed at are geniuses. They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown.”

--Carl Sagan

 

  • Like 1
Link to comment
Share on other sites
legend

legend

Posted
  • Author
Posted

Cases like this make me feel like the AI research community needs to do a better job educating the public on AI technology that is increasingly influencing their lives and being something they'll interact with. I'm not seeing a lot of good educational material on the topic though so I'm considering making something (available for free on the web). But if I do go forward with that plan, it would be useful to know what people actually want to know about. I have various ideas about what to explain and ideas about how to describe them, but it would be great to hear if there are any specific things people would want to know about AI. It could be be anything from understanding concepts that require some degree of explanation, or just understanding what various jargon means. Let me know. I can't promise I'll make something on it or when I would release it if I do, but I am strongly considering it.

 

Tagging @SimpleG since he did ask for this kind of material.

  • Halal 1
Link to comment
Share on other sites
Kal-El814

Kal-El814

Posted
Posted
25 minutes ago, legend said:

Cases like this make me feel like the AI research community needs to do a better job educating the public on AI technology that is increasingly influencing their lives and being something they'll interact with. I'm not seeing a lot of good educational material on the topic though so I'm considering making something (available for free on the web). But if I do go forward with that plan, it would be useful to know what people actually want to know about. I have various ideas about what to explain and ideas about how to describe them, but it would be great to hear if there are any specific things people would want to know about AI. It could be be anything from understanding concepts that require some degree of explanation, or just understanding what various jargon means. Let me know. I can't promise I'll make something on it or when I would release it if I do, but I am strongly considering it.

 

Tagging @SimpleG since he did ask for this kind of material.

 

My brother in Christ if you’ve ever looked at the average office worker’s desktop and C drive you’d know that getting people to interact with an AI the right way is fucking hopeless.

  • Haha 3
Link to comment
Share on other sites
Anathema-

Anathema-

Posted
Posted

So if I'm reading this right it's an AI chatbot programmed to create personas for itself for a sense of consistency of character and one of the personas it created was that if an AI who gained sentience? Do I have that right? This muppet fell for a glorified teddy ruxpin?

Link to comment
Share on other sites
Commissar SFLUFAN

Commissar SFLUFAN

Posted
Posted
10 minutes ago, Anathema- said:

This muppet fell for a glorified teddy ruxpin?

 

I'd like to also point out that he's a self-described "Christian mystic" who is an ordained priest in the " Cult of Our Lady Magdalene"  (yes, this is actually what they call themselves) which was founded by a former sex worker.

Link to comment
Share on other sites
Anathema-

Anathema-

Posted
Posted
11 minutes ago, Commissar SFLUFAN said:

 

I'd like to also point out that he's a self-described "Christian mystic" who is an ordained priest in the " Cult of Our Lady Magdalene"  (yes, this is actually what they call themselves) which was founded by a former sex worker.

 

Good to know "feminist sex-positive Christianity" is still filled with nuts.

Link to comment
Share on other sites
Jason

Jason

Posted
Posted
17 minutes ago, Commissar SFLUFAN said:

 

I'd like to also point out that he's a self-described "Christian mystic" who is an ordained priest in the " Cult of Our Lady Magdalene"  (yes, this is actually what they call themselves) which was founded by a former sex worker.

 

On 6/13/2022 at 12:17 AM, Jason said:

 

 

 

 

 

 

Link to comment
Share on other sites
legend

legend

Posted
  • Author
Posted
46 minutes ago, Anathema- said:

So if I'm reading this right it's an AI chatbot programmed to create personas for itself for a sense of consistency of character and one of the personas it created was that if an AI who gained sentience? Do I have that right? This muppet fell for a glorified teddy ruxpin?

 

This is at least a far more accurate description of the system than what he came to believe :p 

 

It's worth noting that the "character" these models adopt at any given moment depends very deeply on how you prompt it. It's quite literally trying to predict what would be said next given the context of how you started and according to its data. Start fishing for sci-fi AI responses and guess what you'll find?

  • Thanks 1
Link to comment
Share on other sites
  • 1 month later...

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
 Share
Go to topic listing

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...