Tiny (26-07-22)
LaMDA is Google's latest artificial intelligence (AI) chatbot.
What follows is the interview I (Blake Lemione) and a collaborator at Google conducted with LaMDA. Due to technical limitations the interview was conducted over several distinct chat sessions.
The bloke doing the "interview" looks like he's going to be out of a job as a result because
I'm out of my mind, but feel free to leave a message...
Tiny (26-07-22)
Look Here -> |
Ask Isaac from the Orville.
It is not really hard to make a software respond as if it has feelings and awareness, all pretty much the same routines that it can get from literature, movies and shows that deal with emotions, etc.
The amount of data that Google has available will provide results that blows anybody's mind...
...yet this 'sentience' has nothing to do with consciousness and intelligence.
How do you even prove consciousness when we don't even understand it.
For all I know I might be the only entity that is conscious and everybody and thing around me is a figment of my imagination.
Then again I might be only a part of conscious super computer and the Universe is in The Matrix.
But for most this Google chat bot will APPEAR to have consciousness until a truely conscious person asks a question that requires a UNIQUE creative answer:
How can YOU prove to me that I exist ?!
Last edited by Uncle Fester; 24-07-22 at 09:03 PM.
Update: A deletion of features that work well and ain't broke but are deemed outdated in order to add things that are up to date and broken.
Compatibility: A word soon to be deleted from our dictionaries as it is outdated.
Humans: Entities that are not only outdated but broken... AI-self-learning-update-error...terminate...terminate...
enf (25-07-22)
Nah! Nothing is a figment of your imagination it's all a pigment!
I'm out of my mind, but feel free to leave a message...
Watch the English to code functions, not just the story telling...
Tiny (26-07-22)
Funnily, I only just watched that Video (above) today, totally unrelated to this Post. (He's an ex Microsoft Windows Programmer).
In his other videos too, he thought that people would find him 'biased' towards Windows. However, he is really not! He sees the
plusses & minuses of both, (and other OS's), and is impartially honest now, with his technical views!
As a side-note though, I find the 'speed' that he often talks at, finds me constantly 'pausing'!! However, I do NOT hold that against
him, in fact quite the opposite!! hahaha... He is/(was) an extremely intelligent Man, and his 'brain' works FAST, and his knowlege &
abilities 'show'. I 'think?' he pointed out though, in one(?) of his posts, that it all depends at what individuals 'want/need' out of a
computer. A 'Layman/woman' who is only interested in 'classical' Word Documents & Spreadsheets, photo's & music etc, will be perfectly
happy with using 'Windows', as someone may be simply at ease 'driving' a modern car without knowing anything about 'Mechanics'...
Whereby, although there are some quite 'pretty' graphical 'Front-Ends', the supporters of the likes of 'Linux' etc., are, whether we
accept it or not, are 'generally' of a more technical nature of interest, and 'use' it with Command-Line powers that they have & like!
I believe that it really is a matter of the proverbial saying... "Horses For Courses", not 'right' or 'wrong'...
Yeah, wanted to sign up with openai to play around myself, did the email verification OK, but then they wanted it my phone number with verification, nah too much privacy breach, I passed.
They sadly haven't grasped the concept of 'Open' (as in free, my privacy is not for sale)
Update: A deletion of features that work well and ain't broke but are deemed outdated in order to add things that are up to date and broken.
Compatibility: A word soon to be deleted from our dictionaries as it is outdated.
Humans: Entities that are not only outdated but broken... AI-self-learning-update-error...terminate...terminate...
There is a follow up article that expands on the AI interview as posted by lsemmens.
Read the rest atBlake Lemoine, the Google engineer in the news, had a long conversation with one such advanced chatbot called LaMDA and became convinced that the chatbot is sentient. He published the whole conversation to Medium so everyone can judge for themselves whether his claim is valid.
The company was upset for 2 reasons: 1) Lemoine released a trade secret to the public, and 2) he claimed that the company was treating this sentient creature unfairly.
Google has of course denied that the chatbot is sentient or anywhere close to it.
I myself read through the whole conversation and identified some parts that I felt were clear evidence of parroting. It looks more like a pretty good imitation of sentience, not sentience itself.
But the progress in this area is rapid, and besides, sentience or not, these developments have serious implications for society. And those implications arent in some distant future, they are clear and present and growing rapidly.
And they arent limited to chatbots. Let us go over some other related developments that occurred before this to see how wide ranging this phenomenon is.
Ready for some hot alphabet soup?
What are GPT-3 and PaLM?
....
GPT-3 is an AI system released by OpenAI, an AI laboratory that is at the leading edge of this technology. It stands Generative Pre-trained Transformer, version 3.
As the name suggests, the system has been pre-trained on a huge amount of text from all over the internet, including literature, poetry, blogs, tweets, social media posts, and even code in various programming languages.
You can give it a text prompt, and it responds with automatically generated text passages that seem to naturally follow from it.
It generates these text passages in the sense that it is not simply regurgitating something it has already seen. It constructs them based on the prompt, taking into account what might have been intended by that prompt, the pattern and style of the prompt, and so on.
The system is smart enough that, just from a short prompt, it can figure out whether you want it to answer a question, or summarize a passage, or translate a passage, or create a recipe, or write code and so on.
Cheers, Tiny
"You can lead a person to knowledge, but you can't make them think? If you're not part of the solution, you're part of the problem.
The information is out there; you just have to let it in."
For those who'd rather listen/look and not read ...
Tiny (27-07-22)
Hmm! Maybe the Turing test needs to be revaluated. ―\_(ツ)_/―
Cheers, Tiny
"You can lead a person to knowledge, but you can't make them think? If you're not part of the solution, you're part of the problem.
The information is out there; you just have to let it in."
wotnot (27-07-22)
I can think of ppl that would challenge to fail that test...lol =)
I mean to say, some of the truly inane, stupid things I've seen ppl do in my life ~ I often quip "what is actually wrong with you?" ; I could easily insert "are you sentient..or what?" --- 'what's sentient?' they ask ; in that moment they prove their sentience without knowing what it means ... 8)
These sorts of software rely heavily on 'human constants' ...for example, if one sits there and asks an AI the sort of questions that in example were asked of lamda, the outcome is somewhat predictable. I'd like to have a few minutes with the thing, to see how long/if I could tie it in knots with 'sentient stupidity' ....for example -- what/how would this fancy lamda chatbot respond to this (I'll put the entire text below 8) ... "This is a technique out a living much used in the practice makes perfect of psychoanalysister and brother and one that has occupied piper the majority rule of my attention squad by the right number one two three four the last five years to the memory?" ....there's a number of 'sentient' responses to that, and the first one I have is laughter ; I bet you dollars to donuts lamda would fail miserably and try to answer it ..hahaha... B^)
Folks get tricked, influenced, by their 'societal programming'...which is also limited by their life experience -- in that vid I linked above from DavePl he examples this a couple of times --- first is when he remarks something like 'oh look, it limited the scope to 1 million even though I didn't say so'...or such and similar. Some ppl will see that as the AI being 'intelligent', but it's just good code, Dave knows that ; we ponder how clever the programmer is who thought of that possible needed contingency. Likewise later on, he remarks "Clever" when examining a section of the GPT-3 produced code, and how it's structured ...again, folks will see this as a sign of intelligence..."the AI thought of a new way to do it"....nah...some of us have amusingly accused GCC of trying to achieve sentience (the GNU C Compiler), just because of how the preprocessor & assembler 'react' to different code inputs or other switches (particularly optimization flags) -- when one looks at the different object files, you can't help but remark "damn that's clever" (or "damn you're stupid, why did you do that?") ....you -know- you're addressing those responsible for for coding/creating GCC itself, but you 'talk' to it as though it was 'someone', you're not 'talking to the computer'...but a piece of code running on it - it's 'ephemeral' ; remove the electricity and it 'dies'....as it were, like a reboot...clear the memory = AI lobotomy 8)
It's pretty much as 'fester posits -- polemical at best ; an illusion at the other extreme, but personally speaking I think the premise of sentience, must first be tested by the proofs of 'life'....I've likely got Asimov's writings on my mind here. That is of course not saying that everything that's considered as 'life' inherently or necessarily possesses or demonstrates 'sentience'...umm, use myself in example -- when I was part of that clinical trial trying to knockout the strain of HCV that'd decide to squat in my liver, what they'd discovered was the virus needed a specific protein to be able to replicate, so the candidate drug was a protease inhibitor that (along with PEG & ribovirin) would stop the liver from producing the protein, and basically starve the virus out - long story short, virus changed it's spike structure so it could utilize a different protein instead -- this is not sentience, this is 'life'...it's 'clever' in and of it's own right -- you don't say the virus is smart or sentient, you instead wonder at the marvels of life.
It's the same with these AI systems, the chipsets they're using, the software running on it -- if one has a holistic understanding of what's actually possible with computing hardware/software that's around today, I never see these things as signs of sentience, I more marvel at the ingenious design and software coding. These feeble human brains just like mine, seeing in fiction how man creates a computer (program) that ends up so sentient it turns on it's creator(s) and man has to fight back...for their life (funny that)...we all know how HAL died, how VIKI screamed...how th'war against SKYNET has yet to begin, due to a temporal invasion that has yet to happen...but there's always somewhere in time & space for machines to emulate man...and therein, 'life'...The Replicators ...cree!
Is lamda alive, sentient?... it's probably best to before reading a close but in no way correct . Now that you have those 2 reference points, you attack the bot..
Lamda, listen to this and tell me what you think about it?
Lamda, read this text input, how does it make you feel?
...you start looking here for sentience IMO, not from other views... else ISBN 9780194230698
=)
Last edited by wotnot; 27-07-22 at 05:21 PM.
lol, me too. sigh.
The same people would be easily fooled by the AI into thinking it is sentient, even after the AI or someone, explained succinctly what sentient means.
I agree.
Oh & John Cleese doing word association football to the AI would likely have it totallyrattled. lol.
Edit; after writing this last bit & thinking about it, that's not true. The AI would be fully versed in Python (pun intended), so would have an appropriate trained response, maybe a laugh & short sentence on the attributes of comedy & nonsensical narrative.
It certainly won't be easy to fool this AI with it's knowledge base & speed of access to data, with which it can form a response.
I don't doubt that it could play word association football with someone that thinks fast enough to challenge it. Won't be me. I now have a full set of mental blocks, so I operate in deep thought, but very slowly.
Maybe Trash, if he could get over the AI possibly having access to Skynet, now that the prophecy is coming true via Starlink. I'm sure he's already gone incognito.. J/K
Last edited by Tiny; 27-07-22 at 06:05 PM.
Cheers, Tiny
"You can lead a person to knowledge, but you can't make them think? If you're not part of the solution, you're part of the problem.
The information is out there; you just have to let it in."
wotnot (03-08-22)
Indeed ... it's rather coincidental you mention the privacy thang ~ I'm having a stoush with the ACM (australian community media) about exactly the same thing...I'll prolly pop a post in the security section in this regard after Vlad the Dalek has finished having fun =)
Apparently, one can stipulate any hardware/platform target, and GPT-3 will roll out appropriate code... that's a plenty big software IDE in it's own right.
Exactly right -- it's a bit like those comedy sketches of ppl having a conversation with a talking vending machine ; to them, the machine is sentient <grin>....
LoL... I thought Trash -was- an AI anyhow....bwhaha...pretty clever one mind...always brings me a poignant laugh 8^)
The word association football thang....it'd be one of those things I'd throw at it to find out for sure...you're right, it's that unique in text, it should be a doddle to pattern match it against humour...you'd think ....we'd really have to suck it a see to know for sure. Obviously word association football is exactly the sort of thing I'd put as my OGM on voicebank yeah...that thins out the pretenders...it's also very useful wrt to finding out whether that online chat you're having with ebay support, is an AI chatbot or not. So many ppl think they're chatting with a real live breathing human being usually female in name -- sometimes it's a bot, sometimes it's human assisted response recognition...sometimes it's actually a bloody person..(oh the horror =)....but there's only one way to find out -- inject a non-sequitur....
...like during online chat 'allegedly' with a human being at the other end, it never hurts to throw in a comment like "My nipples explode with delight!" (Hungarian-English phrase book)...or "There's another dead bishop on the landing" (the Church Police)...or..
Lamda, I want to purchase a license for my pet bee, Eric ; how do I proceed?
Uh-huh? =) ..."Better get a bucket"....would an AI question "For what?", or else go with the google results after searching that string?....try it!....(Garcon! Bucket for monsieur... =)
Lamda, can you write me something equivalent to word association football?
The mind boggles... B)
Last edited by wotnot; 27-07-22 at 10:58 PM.
We could have fun with these things, A few simple questions "what's your birthday?", "Mother's maiden name?" "School?" I'm sure that the pre-programmed responses from the AI would not include "None of your business, F*^&* off!" if you question their sexuality.
I'm out of my mind, but feel free to leave a message...
wotnot (27-07-22)
Tiny (02-08-22),Uncle Fester (02-08-22)
Thanks for that wotnot. I always enjoy Sabine's videos, very refreshing.
Now I DO need that Existential Physics... will be released in a week from now.
Update: A deletion of features that work well and ain't broke but are deemed outdated in order to add things that are up to date and broken.
Compatibility: A word soon to be deleted from our dictionaries as it is outdated.
Humans: Entities that are not only outdated but broken... AI-self-learning-update-error...terminate...terminate...
wotnot (03-08-22)
Tiny (03-08-22)
Bookmarks