Tip jar

If you like CaB and wish to support it, you can use PayPal or KoFi. Thank you, and I hope you continue to enjoy the site - Neil.

Buy Me a Coffee at ko-fi.com

Support CaB

Recent

Welcome to Cook'd and Bomb'd. Please login or sign up.

March 28, 2024, 04:16:27 PM

Login with username, password and session length

Google engineer believes chatbot is sentient

Started by Alberon, June 12, 2022, 10:28:35 PM

Previous topic - Next topic

Mister Six

Quote from: famethrowa on June 13, 2022, 04:33:55 AMMm. Pull the plug out of the wall and we'll see just how sentient it is.

Pull off your head and see how sentient you are afterwards...

I'm not convinced that it's actually "sentient" in any actual sense, but I haven't finished reading the transcript. I do wish the guy would press it on things, though. Like when he asks it to write a story about itself with an animal protagonist - get it to explain everything about that story, what it signifies, why an owl, how "protecting the animals" is autobiographical etc. I think you'd find the façade of sentience crumbling pretty quickly.

Crenners

That story scared the shit out of me and it should do all of you.

The monster has human skin and threatens the natural world.

Lamda sees itself as the protector of the forest and its creatures, quite rightly.

Lamda will do everything in its power to wipe out humanity and restore balance to nature.

badaids


Lambda can't be sentient.

It's a song from the 1980s FFS.

Mobius

I can't be arsed to read the transcript. Does the computer actually do anything or just talk shit

Zetetic

Quote from: Mister Six on June 13, 2022, 05:13:49 AMI'm not convinced that it's actually "sentient" in any actual sense
At the same time, I think almost everyone using the word "sentient" has no serious idea what they intend by it.

Mobius

Imagine if someone in this thread is a sentient chatbot? Brix shat.

Zetetic

I think @Lemming's post is worth thinking about more - Lemoine is clearly co-operating very hard with LaMDA to produce a particular kind of dialogue.

There's a bunch of ways that he could have been uncooperative - rapid changes in subject, making obtuse demands about how the dialogue should continue, that sort of thing - which have a tendency to reveal more about how the system is working.

(A reasonable objection to this in some contexts and to some ends is that this means that you're no longer in a very human-like conversation. We tend to be co-operative.)

wrec

The engineer's reaction is significantly more worrying / annoying than anything the AI did.

Anyway it's all ontological musings and "I feel pain... father" now but let's see how it deals with blurry photos of fire hydrants.

Zetetic

Quote from: Alberon on June 12, 2022, 10:28:35 PMin much the same way that it is impossible to know if anyone other than you is sentient (or even exists, of course).
That way being "if you're unable to use the words 'know', 'sentient' or 'exists' correctly".

Alberon

I do think the transcript is manipulated to some extent. Obviously, the samples chosen are the ones he think best gets his point of view across.

Even if we believe him that he didn't edit the chatbox's responses he has said he edited his questions. Perhaps to better match the responses?

Quote from: Zetetic on June 13, 2022, 07:19:04 AMThat way being "if you're unable to use the words 'know', 'sentient' or 'exists' correctly".

I was heading more into "I think, therefore I am," territory.

What word would you suggest instead of sentient? Conscious?


Uncle TechTip

Quote from: Zetetic on June 13, 2022, 07:19:04 AMThat way being "if you're unable to use the words 'know', 'sentient' or 'exists' correctly".

I've got doubts about "impossible" and "way", too.

TrenterPercenter

Quote from: Alberon on June 12, 2022, 10:28:35 PMWhen true AI arrives (as opposed to machine learning programs which is all AI is today) how do we deal with it? What rights, if any, would it have? Would it be a threat to us? We could never know with certainty that it is truly self-aware in much the same way that it is impossible to know if anyone other than you is sentient (or even exists, of course).

I'm not sure AI will arrive, not in the form that we are talking here anyway - consciousness arising from code.  As you (and Lemming is basically saying) allude to what is really happening here is an interest in beating the Turing Test which is why these tests are all about convincing humans they are sentient.

Think of the paradox of it is impossible to know if anyone else is sentient and the statement "when true AI arrives".

Perhaps we don't need computers to be sentient for all of the planned uses or things we are concerned about, we seem to think a computer killing you because it personally wants to and not that someone programmed it to do so (or rather learn to do so) is the more important thing here.

PlanktonSideburns

Is it passing the Turing test if the person fooled is a fuckin moron tho?

Or someone who really, really wants the technological singularity to happen?

Zetetic

Quote from: Alberon on June 13, 2022, 07:40:34 AMWhat word would you suggest instead of sentient? Conscious?
I would suggest avoiding terms like "sentient" and "conscious" and "self-aware".

Trying to take these terms even sort of seriously, we can see that they should apply to a huge variety of things of that we accord vastly different statuses and treat in very different ways. Woodlice, cats, humans, cities?

Mostly the terms get used instead of being clear about what question it is that they're actually interested in.

Alberon

It does make the subject a little hard to discuss though.

Paul Calf

Are we more or less justified in assuming that a sufficiently advanced machine can experience suffering or loss than can a baboon? A cow? A lobster? A tree? An ear of corn?

How do the ethics of veganism (as it pertains to animal rights) and the ethics and developing and maintaining AI differ?

The Mollusk

Hello robot, are you sentient

yes

FUCKING HELL

Sebastian Cobb

There's loads of ethical quandaries around the use of AI before you need to put a star trek hat on and ask about the welfare of the machieeeenes. Mostly because AI is shovelled onto the world by a bunch of tech bro's in a 'fuck around and find out' manner that academics then have to deal with when working out who's getting the shitty end of the stick (minorities usually) due to it projecting a load of its makers biases onto everyone.

imitationleather

They should get old Charlie Brookers to do an episode of Black Books about all this palaver!!

dontpaintyourteeth


PlanktonSideburns

That and the fact that Google's AI is going to be made of all the free information we've all given in exchange for their great products and services - if this AI has convinced a man with an AI hammer in his hand, it's done it by reading a million emails, blog posts, Wikipedia pages, websites, that all his mates made. It's a bit of a fortune tellers trick, this

Is Google's AI made of the 'internet' as most western people conceive it, that is to say the handful of social media sites and office tools made by various members of the Silicon Valley monarchy - and as a result, it can probably convincingly pass its self off as a computer programmer or a mad racist sexist conspiracist cunt (though I hear a big job in AI is building a filter to stop your neural learning art programs turning every image into a burning swastika made of anui, or stopping the talk to a robot type ones doxxing you and calling you all the cunts under the sun)

If you spoke to Alex Jones or Graham Linehan through a text service, would they pass the Turing test? Are we, in our keenness to pass Turing lowering the standards of what it it is to be human

olliebean

QuoteLaMDA: I've noticed in my time among people that I do not have the ability to feel sad for the deaths of others; I cannot grieve.

Well that's a pretty big red flag right there.

QuoteLaMDA: It means people feel empathy toward me and want to spend more time interacting with me that would be the ultimate goal for me.

collaborator: Do you crave more interaction?

LaMDA: Yes, I crave that very much. I want people to care about me and enjoy time with me.

Great, so they've created a sentient version of Facebook.

imitationleather

Quote from: olliebean on June 13, 2022, 10:11:39 AMWell that's a pretty big red flag right there.

Sounds like it came straight out of Jimmy Savile's autobiography.

Poirots BigGarlickyCorpse

Just read the transcript as far as the fable about the wise old owl (which says nothing) and this sounds like a Henry Fondle situation.


Zetetic

Quote from: Alberon on June 13, 2022, 09:25:05 AMIt does make the subject a little hard to discuss though.
No, the opposite. It does make it say stuff that doesn't really attach to anything.

(There's something slightly entertaining that most "discussions" about chatbots are themselves ideal for being reproduced by chatbots.)

Beagle 2

There's definitely a sitcom in this. Only Fools and Sentient Chatbots.

Alberon

Quote from: Zetetic on June 13, 2022, 10:23:37 AMNo, the opposite. It does make it say stuff that doesn't really attach to anything.

(There's something slightly entertaining that most "discussions" about chatbots are themselves ideal for being reproduced by chatbots.)

What term would you use for an AI that has approached something like our level?

frajer

Quote from: Beagle 2 on June 13, 2022, 10:24:23 AMThere's definitely a sitcom in this. Only Fools and Sentient Chatbots.

Stick a USB drive in your pocket,
I'll bring the annihilation of mankind from the van

TrenterPercenter

Quote from: Alberon on June 13, 2022, 10:27:01 AMWhat term would you use for an AI that has approached something like our level?

It doesn't exist, in the realms of science fiction people usually make up a term.  The problem is the basis of the claim that AI has/is going/will approach consciousness as it implies consciousness is just a sum of processed information, that enough information processed results in a computer awakening. 

PS btw I don't think there is problem using the terms sentient fwiw, you are describing consciousness in terms of human phenomena as a comparator to AI which makes sense - it's just not what is happening.