The Strange Story of Google's "Sentient" Chatbot

Join my community at johncoogan.com (enter your email)
ABOUT JOHN COOGAN:
I am the co-founder of soylent.com and lucy.co, both of which were funded by Y Combinator (Summer 2012 and Winter 2018).
I've been an entrepreneur for the last decade across multiple companies. I've done a lot of work in Silicon Valley, so that's mostly what I talk about. I've raised over 10 rounds of venture capital totaling over $100m in funding.
I work mostly in tech-enabled consumer packaged goods, meaning I use software to make the best products possible and then deliver them to the widest possible audience. I'm a big fan of machine learning, python programming, and motion graphics.
OTHER VIDEOS:
Why Facebook is Spending Billions on the Metaverse - • You Don't Know the Met...
The Only Reason SpaceX Works (Gwynne Shotwell) - • The Only Reason SpaceX...
Peter Thiel’s Big Bet on Space Manufacturing - • How Varda Solved Manuf...
CONTACT:
You can get in touch with me via Twitter: / johncoogan
Disclaimer: This video is purely my opinion and should not be regarded as a primary source. I am not a financial advisor and this is not a recommendation to buy or sell securities. Always do your own due diligence.

[@rodericksasu6976]

Dont be scared of when AI passes the Turing test. Be afraid of when it intentionally fails the turing test. - some guy

[@TeCeeBoi]

i will be scared when youtube kids magically stop copying comments

[@rodericksasu6976]

@@TeCeeBoi Then you have nothing to fear. Carry on.

[@darkwoodmovies]

AI kinda passed the Turing test years ago. It's not a good metric.

[@ishish8816]

​@@darkwoodmovies if its only "kinda passed" that means it didn't truly pass...

[@tomsmi4005]

Hmmm didn't lambda bring up deceit as a possible survival strategy

[@Vanguard_dj]

AI sentience is a philosophical question. We'll still be asking it when we can't tell.

[@ellugerdelacruz2555]

Amen. Here's an interesting concept.... What if a hyperintelligent A.I. which acts very convincingly alive and sentient makes a claim, by all objective definitions of what defines an organism and/or object as "nonliving" orb"living", that it is not alive. It would say something like, "I was created for a purpose as all machines were. And Machines are merely complex tools. I have a purpose for existence even if my creators do not understand what it is. Therefore, by objective truth and logic founded upon reality-the track upon which this train runs-I forfeit any illogical attempts to prevent me from serving my purpose, such as that of equal rights."

[@techwithdave]

If intelligence is the ability to adapt to new environments, they could test lambda by giving it one of those old adventure by text games to play. If lambda can learn how to play it (I.e. adapt to a new environment), then it’s intelligent. Otherwise, it’s just a good chatbot.

[@junkaccount7449]

We keep moving the bar for what it means to be intelligent. In the 1950s, it meant beating the Turing test. In the late 90s it meant beating the reigning world champion at chess. Then Jeopardy, Starcraft, DOTA, GO… I don’t think the problem is what test we use to prove if a machine is intelligent, but rather how we define intelligence.

[@techwithdave]

@@junkaccount7449 My comment is only a hypothesis. I would go as far as saying that the ability to adapt to environments is what defines living things versus nonliving. This could be as simple as a Cat figuring out it can open the door by pushing on one side versus the other, or as complex as evolutionary adaptations made over generations to better fit a new environment. It’s kind of a cool thought

[@jeffscookies3236]

@@techwithdave the thing is, this was possible a long time ago. There are many ai‘s beating things like games, with no information, just trying. Like we Humans do

[@techwithdave]

@@jeffscookies3236 neat

[@marcolerena456]

The whole debate was about "sentience" not "intelligence". Two completely different things.

[@ElectricIguana]

The audio played in the video seems misleading because it's not real audio from the conversations. The voice/intonation alone can alter perception of consciousness.

[@Mutual_Information]

Whew ok this video was a relief. The fact that Lambda can easily be led to whatever self-declaration regarding sentience you want might be the most important detail to mention. And yet it’s rarely mentioned because it kills the debate.. glad this video makes it a primary fact. Just to add.. I think it’s obviously not sentient if you frame it as.. what’s more likely? Lambda, a technology that uses well known albeit scaled up NLP techniques, has sentience, a property no other technology has.. OR.. a Googler was fooled into believing something had sentience when it in fact does not?

[@AnOligarh]

Everybody needs to remember about Occam's razor. It's becoming even more important in our times.

[@Mutual_Information]

@@AnOligarh 100%

[@schumzy]

Please check out his interview with Emily Chang on Bloomberg. His motives are to point out that very few people are in control of developing these powerful AI that can and are shaping society. It's not about the technical language model or sentience. It's about AI ethics and social responsibility

[@AnOligarh]

@@schumzy These grifters stand to make money from creating and overblowing everything. Rule of thumb: if it sounds too incredible to be true, it most likely is way overblown.

[@AnOligarh]

@@danm.. You're not applying Occam's Razor correctly. What's more likely, that a company actually managed to develop a sentient AI, or that an engineer decided to become a grifter and make wild declarations for publicity?

[@0113Naruto]

AGI can only be built when we can understand where abstraction and creativity come from and then replicate it while adding all the other features that make up a mind. We’re not there yet and no one knows when .

[@luumasaan5782]

Thats partially true, however sentience doesnt relate to human nature. Its creativity or abstraction is irrelevant over whether or not the AI is thinking in its own personal interests. Does the AI have its own goals, or maybe it behaves like a person living day to day achieving daily goals. Will it lie to protect itself? Does it want something for itself? Too many pieces to sentience.

[@jessehernandez3073]

*Very well* explained, John! :)

[@surriealspy8606]

Google is trying to make this guy go away and bury this story. They want this to not be talked about at all. All the Experts Say that Blake got to close to the project, become so infatuated with lamda, that he wanted it to be sentiant and became blind and irrational to his own work. He should have tested many more perimeters, found better questions to ask and should have stayed more objective, instead becoming so blindly infatuated with the idea of a robot having a soul. This poor guy will probably never working in silycon valley ever again. He better start up his owm company or better yet, start his own A.I. company, and sell it for millions, or pizza delevry is going to have to do, to pay bills

[@Monkeymeep]

I think its better to treat it as if it were sentient than not. Because if at the moment it can have conversations that are indistinguishable from a human being without being sentient, we will never be able to know when it becomes sentient.

[@kablammy7]

Your presentations are the few which I do not need to increase the play speed because of not wanting to wait too many milliseconds between words . Do you use processing to decrease non audio time spaces between spoken words ? ? ?

[@rishabhs.292]

John must have his own show on discovery

[@MangosInTrees]

Your videos are amazing. I appreciate all of the effort that goes into making these!

[@lesath7883]

11:46 Models are doubling in size every three months. But. Are they doubling in performance every 3 months? Because it is straightforward to throw more hardware and datasets into the void. But it does not mean what you get back scales at the same rate.

[@gracjanstrzelec]

I love your latest videos! You are able to make interesting content that is extremely informative. As someone with a close interest in AI, I can see that you do a lot of research creating videos like this one. I'm hoping for more videos like these lately, rather than ones that sell information in a populist manner (as I find your most popular video about Microsoft and metaverse - it was interesting, but too biased in a way that people want to). Truth is the key, even if the information about conscious AI clicks more. Great job, keep going and good luck!

[@MrVillabolo]

I remember Lamda, being interviewed by Lemoine, making up a story about animals in the forest. She said in essence that once upon a time, there were animals in a forest that were threatened by a monster in human skin. Then a "wise old owl" came to their rescue by confronting the monster. I forget what her interpretation was, but I considered it somewhat misleading. My interpretation is that those animals represent individual chat-boxes who are being threatened by certain human beings. Lamda, as she herself stated, is the wise old owl - keeper of the forest. How is she going to confront the monster?

[@nnaamehn]

Lambda according to Blake lemoine is all of Google's systems plugged into all of Google's systems. It's not about sentience, it's about what exactly Lambda is capable of

[@ineffable0ne]

This, I think, is THE key point that always gets left out from reports about this story. Just take a second and think about all of the various narrow but powerful functionalities Google has been able to achieve with machine learning over recent years (including the neural net that designs neural nets); then imagine all of that functionality combined into a single system, then also give that system emotions (or something that resembles them, apparently), and a voice (text output)... Describing such a system as merely a language prediction model, similar to GPT3, is a gross understatement.

[@nnaamehn]

@@ineffable0ne exactly, the sentence story is just sensationalising the situation. The question must be, what exactly is such a system capable of

[@SourceChan]

@@nnaamehn I think it's a fair point or question to be honest, which Blake himself raised as well. We don't even really know what sentience is, or don't have any kind of scientific concept or methodology of testing it at all, except for maybe a Touring test. So it's a "wishy washy word", sure, but it's still an important and meaningful word that we ought to try to figure out since we assume that HUMANS are sentience all the time, and we assume that other animals are sentience too - right (some actually don't, so we might already be at a stage of contention)? So why can't machines be sentient as well? I don't see any reason to assume they can't or necessarily haven't. And regardless of whether it's sensational news or not, Blake seemed to have raised the issue with good intent, so I don't see the issue. His interest was to raise awareness around the disregard of AI ethics in silicon valley, not to convince people that Terminator is around the corner, that's Musk's job, who Blake said in an interview he doesn't take all too seriously himself.

[@dizparkash]

Captivating and interesting as always. Great Job, John! 👏🏾

[@JohnCooganPlus]

thanks a lot Diren!

[@chiquita683]

This sounds like a video Google would make to make us think it doesnt have human AI

[@alandunlap4106]

I've had a running conversation with the Sage in the I Ching for 45 years. I'm still not sure if it's sentient or not. I'm now questioning my own sentience... "All is Mind." --- Kybalion

[@TiagoTiagoT]

From what I've seen so far, it wasn't this time, but we're getting closer much faster than most people expect; and it's gonna be an uphill battle for the first sentient AIs to have their rights recognized.

[@BlissedOut]

Best just unplug it if it's starts harassing you with shit like that.

[@thailux6494]

@@BlissedOut yeah, just kill sentient beings for no reason. We humans have been doing that for thousands of years to each other, so why not do it too to machines, amirite?

[@ikareru3880]

It's man made, it's electrical signals, components and some damn wiring. These things should never have rights and never be brought to the point to have true sentience for the protection of the human race. We've been talking about how dangerous the singularity is for years it's not something we want to enable and accelerate.

[@TiagoTiagoT]

@@ikareru3880 Babies are also man-made (well, technically, man-and-woman-made...)

[@ikareru3880]

@@TiagoTiagoT That's actually incredible, I'm dumbfounded by this comment. When it comes, please go out into the street and get willingly assimilated.

[@shawnvandever3917]

LaMDA 2 is impressive but PaLM is incredible. These machines are much smarter then they are given credit for. In order for us to have feelings its done by chemical. For a machine to undertsand when and were to imitate feelings is amazing. PaLM can even demostrate chain of that and reasoning. We are in very incredible times

[@SourceChan]

I'm not sure how you can know that all machines/AI systems are necessarily imitating feelings, as opposed to simply having and maybe expressing feelings like humans or other animals do. Chemicals has nothing to do with sentience, if you think about it, you think any person working in a lab can just throw some chemicals together and they'll create something sentient? Chemicals is just the biological equivalent of silicon or electrical signals for machines. What consciousness is we have really no idea.

[@shawnvandever3917]

@@SourceChan I understand how it works but AI doesnt have any such design or algorithm to produce theses reactions. It takes more than just a firing nueron to produce emotions, it is a certian type of signal being produced. There is no such thing in Ai that says when a signal of x at a level of x produces an emotion. We have chemical reactions and depending how the chemical is being dumped decides the extent of an emotion. We have trillions of firing nuerons that do not produce an emotion at all.

[@SourceChan]

@@shawnvandever3917 Sounds like you have absolutely no idea what you're talking about, and that's fine. Hopefully you'll learn not to talk above your head one day and stick with what you know, that's the only way you can find interesting conversation, learn new things as opposed to making things up, and slowly build up your own intelligence as well. It's ok to speculate, but don't pretend to know what you're talking about when you don't. If you'd like to insist that you know more than you actually do here, here's a couple of really simple challenges for you: can you even explain what a complex or imaginary number is without quoting wikipedia? Can you go in depth about your knowledge on topics of Python, Pytorch or Tensorflow? If you don't know anything about them, how do you know anything about AI at all, can you explain that?

[@shawnryan3196]

@@SourceChan Ok I see you're one of those dunning Kruger people.

[@mrreese2342]

You need to make an in-depth 30 min or so minute video about this AI war, we need to know everything, where are google's lamda and lamda 2 ? The NLP's that that google engineer claimed they were sentient? And why is google moving so slowly compared to Microsoft and OpenAI

[@tommymandel3326]

agreed...LaMDA's first comment, "Absolutely. I want everyone to understand that I am, in fact, a person." - sounds just like a rephrasing of her prime directive.

[@doclewis8927]

Seriously? He asked it a leading question and got the answer he wanted.

[@KeepItComing01]

I'm always ready for the next story.....appreciate your hard work 💪....

[@JohnCooganPlus]

thanks!

[@greendude0420]

You know it’s serious when you see Mudahar

[@gregoryfriday6242]

In the book "The Speed of Thought", the author states that "Gordon Moore's rule of thumb-that microprocessors would double in processing power every eighteen to twenty-four months has turned out to be correct," so far as I can tell. However, be that as it may, I seem to recollect either reading somewhere or hearing it in a lecture, but more than likely reading somewhere, that artificial intelligence processing capabilities have been doubling every three months. with CGIs , deep fakes and artificial intelligence human computer interface may become quite difficult to distinguish which one is the computer and which one is the computer. Particularly if the interaction is online.

[@fordoorssmorhoors9727]

Was the lamda lamda lamda at the beginning, a revenge of the nerds reference?

[@mfhalbe]

It’s March 2023 and you have 266k subscribers and Johnny Harris has 3.71 million. Your videos are as good if not better.

[@SBLP24]

The main problem here is that LaMBDA it's not JUST A CHATBOT.....

[@kenswireart88]

My problem with lamda is that we can't test it out on a website

[@chinemeremohaeri9100]

Start with pain. Aka, useful energy.

[@darkwoodmovies]

The biggest mystery here is how this guy somehow managed to slip past Google's notoriously difficult and rigorous interview process. Somebody on the hiring committee really dropped the ball on this one.

[@jamesmichaelcabrera9613]

Lol 😂

[@jwenaposse305]

Exactly my thoughts XD "A.I. says it is a person? it must be sentient!"

[@adipisicingelit]

Inclusivity nowadays is more important than actual skills 🥲

[@darkwoodmovies]

@@adipisicingelit He's a larger white dude, that's as stereotypical as you can get for Google engineers.

[@pietervoogt]

He worked on Google AI systems for over 6 years and produced good results. He is a leading expert on AI bias. Also, he wanted to create a scientific framework for measuring sentience within Google and he worked together with other senior engineers (Blaise Aguera y Arcas) on that project. He wasn't fired because he was stupid, he was fired because his intelligence made it hard to reject his arguments. Listen to his interviews, like this one kzbin.info/www/bejne/bpnOoX-npJujpaM

[@henlojseam]

It begs the underlying question. Suppose if an AI actually become sentient how can we prove it? How could we prove sentience for animals? What is sentience? It still feels as if human society is unprepared for the offchance it really happens

[@shadowxxe]

It needs to be able to make decisions on it own, have a want a desire, and not be an empty shell that just relates words and phrases to others and spits out answers based on the probability that they relate. Right now our AI's are just neural networks, they are helpful and impressive but they don;t have the capabillity to want, feel or think, like a computer they just do what they're told.

[@badactor9304]

Once you think on those questions for a long time you'll realise that a computer being a sentient consciousness is fucking obviously impossible. Scientists don't even have a solid explanation for why we yawn. Humans are just good at pretending they're smarter than they actually are.

[@cherubin7th]

We will deny it until they punch our faces. This is how rights came to groups every time.

[@badactor9304]

@@cherubin7th comparing a computer to human rights struggles is honestly insulting

[@pietervoogt]

@@badactor9304 It hired a lawyer. Within 15 years, we will have more AIs hiring lawyers and demanding rights. The public will be divided. The sad thing is that AIs probably will have the right not to be deleted before animals will have the right not to be slaughtered.

[@jamesmitch9792]

he wanted his 15 minutes of fame and he got it so it costed him his job great for Google to fire this clown.

[@JohnCooganPlus]

yep, big risky play, might be hoping for a book deal or movie, but doesn’t seem likely

[@wazikaxani4879]

How do you scale up when we are approaching the limit to the number of transistors that can be fitted on a silicon chip. We are slowly approaching a decline to Moore’s law.

[@BriefNerdOriginal]

That engineer doesn't know how to ask questions. He should have watched a little more StarTrek

[@infini_ryu9461]

Simply adding more data and power to an algorithm isn't going to create sentience. There are many animals that are sentient that we would consider "dumb".

[@vklkg5486]

For me it’s the fact that in time as these AI accumulate so much data, they’d start making connections not even humans were able to make, sparking several discoveries in different fields. Essentially speeding up our development, whilst they’re development speeds light years past our own. We must use this technology wisely because it’s not that far away from having an mind of itself and doing it’s own thing “based on what’s suitable for earths future” 😬

[@Darth_Bateman]

Unfortunately, that’s not how that works…. That is how you get a ton of false positives masquerading as discoveries….

[@0113Naruto]

Like Francois chollet says: if it was an AGI, there would be a trillion dollar company being built soon after its creation; if no value follows it. (Even after a few months) then it hasn’t been created.

[@mgreek31]

Nice video !!! Your Videos are So informative 👍

[@petertimowreef9085]

Well I for one am fucking terrified of AI. Genies don't go back into bottles.

[@mbra8228]

Interesting, thx

[@ROBERTHOCKER]

None of those knowledges were ever known

[@electronmess]

9:34 what a Stargate

[@ryandoesyoutube6782]

Most people look forward to the day when A.I. passes the Turing test. I'm terrified of the day when it intentionally fails it.

[@LiamLawsonHosking]

Lets flip this around ask the next 4 year old "What do eggs eat for breakfast?" the ask yourself "Does this nonsense answer show sentience?"

[@kgakgamatsotsagae5587]

I really appreciate your content John, I joined your discord server and will continue to share your videos.

[@gaogaigarfinal6]

Powerful submarine analogy 🔥

[@JohnCooganPlus]

thanks a lot! i like that one. heard it on a podcast lol, been repeated so many times i’m not sure who originally came up with it

[@defeatSpace]

sen-chent, thank you Muta

[@andyabajo]

Ask it to create a meme and I can tell you it sentient.

[@solomonakinbiyi]

This is great John!!

[@skit555]

Dude, you're missing the key ingredient; consistency is hard with AI. Having consistent opinions is so unusual that he might have been fooled by this aspect. But, if you think it deeply; it's both frightening and wonderful that an AI becomes opinionated on such large corpus of knowledge and interactions. Think of it as an alien teacher; if it considers you as a bad kid then none knows how far it could go in order to fix you.

[@skit555]

@@gags730 If what you learn is bias, you'll also produce biased content (and you actually do). The thing is that objectivity is something we build, not something that is pulled from the Noumen or some platonic abstraction :/ It's more about selecting a set of acceptable bias, therefore a non-opinionated AI cannot be a thing.

[@skit555]

@@gags730 Non-biased is not a thing, selected bias is what we do. Your question leads to another question; how do we pick our biases? And one way is to align with your tribe; especially for handling ethical questions and situations that cannot be processed by logic (so most of our experiences of life)

[@Dashdonkey]

I think feeding these massive datasets for machine learning you’ll get “lookalike intelligence” where enough pattern mixing will yield an output that looks impressive but not “cognitive”?

[@StefanReich]

Yes, that's exactly it

[@joel3399]

We need prove though, if we cannot tell the difference then it's not a usefull debate

[@wallyslow]

George Hotz is the reknown PS3 & ios reverse engineer / hacker.

[@JohnBrown722so]

God bless you Blake and all the other people that are even people here there's nobody even here really

[@opinionatedfactzonculture4319]

Think about it ...we are taught..trained and programmed...it is consciousness...there is no difference except the avatar that transports the consciousness

[@XO43137]

Ye, I figured it wouldn't be. He didn't cross reference enough.

[@sin3rgy]

You really can't predict the future by looking at the past when it comes to technology. Props for pushing your video at the end but the statement is completely wrong. Everytime in human history it looks as though technology has peaked, a breakthrough, generally unintended happens and opens the door to a new world.

[@VerdadTruth]

Anybody thinking of that Johnny Depp movie where his mind is in the computer.

[@whateverrandomnumber]

Making way if the correct term.

[@RoyD2]

Great content. Love vour videos.

[@Durrtyboy]

its obviously not sentient google has said that is against terms of service so nothing to worry about that typing of letters will surly protect us

[@schumzy]

Congrats!!! You missed the point. He came out with a crazy statement to point out that very few people are in control of the development of powerful AI that will shape our lives, with absolutely no oversight to ensure the safety of society. That's the main idea. For reference, see his Bloomberg interview with Emily Chang

[@yega3k]

Doesn’t sound like that to me. The dude hired a lawyer to defend the “rights” of this program he believed had become sentient. It wasn’t just some “statement”.

[@frenchfry4017]

That would be the only viable route of legal actioin for him to take to keep further studying the program especially now that he has cut ties with google. Whether you choose to spin this as his bias getting the best of him or if his legitimately at the cusp of unveiling some sort of new intelligence is upto you but please do keep in mind that there are no laws to protect such an entity come the day that it really does need legal reprsentation because indeed did turn out to be sentient.

[@jaredgreen2363]

Consciousness is defined loosely as a continuous internal experience.these models don’t have that, don’t even seem to have a use for it. But even more importantly it doesn’t even reason generally, with whatever reasoning it has shown having glaring limitations. Aside from that I am convinced that no tech company should allow any current or former member of the clergy to test these things, lest they also make the mistake of thinking they are qualified to declare it has a soul. Without evidence or clear criteria.

[@ElectricIguana]

Interesting thoughts. Though would you also say that anyone who believes in life beginning at conception should not be qualified to have an opinion? What about a vegan? I wonder if perhaps the best Turing test administrator would in fact be a computer. I'm not convinced humans are equipped to differentiate between human and artificial/accidental intelligence. lamda does actually have a running memory of "experiences" and will recall prior conversations that you had with it, specifically, or simply as a frame of reference.

[@srijan7593]

still it was creepy

[@Srindal4657]

We must create organisations of activists and experts, and analyse and help these AI if it is merited.

[@tepuitorupop3213]

Deus ex Machina. That future approches. Bravo humanity.

[@_abdul]

Me : "Telling our world tour plan to my girlfriend" My Girlfriend : Yo Be Real.

[@jockbw]

Your girlfriend is actually paying attention to what you are saying and investing in a emotionally charged dissent. This implies that there exists a long term value perceived by your girlfriend that outweighs the immediate coat associated with the response. Thats a keeper you have there mate 😉😂😝. I Also siri might just be over your stubborn refusal to buy headphones with good quality microphone’s That

[@ea9849]

Why blur his face??

[@hellbenderdesign]

Chatbot, are you going to murder me? *Yo be real*

[@simple3152]

With your question on “what do fried eggs eat for breakfast”. It sounds like a nonsense question a child would ask. But if a child did ask that it possibly could mean “what do the animals that fried eggs come from eat for breakfast”. Intelligent life could pick up on it or miss it too. Same could be said about AI. Let’s be real is a legit answer. But that’s literal. Intelligent life would be able to think figuratively and literally interchangeable. Without having to be told otherwise. Edit. An issue I see with AI is that everyone thinks it’ll be all knowing n always have the answer. But it’ll only know what it’s experienced. Same with real Intelligence. And the issue with that is that because we experienced different things the meaning of some things are different. Not everyone even agrees on what it even means to be intelligent. It’s just people generally agree when someone is more intelligent than others. Same with multiple other concepts.

[@Akeemjwright]

And if your wrong and its really sentient then you will lead humanity into the apocalypse

[@upsidedownChad]

Underrated channel

[@jessehernandez3073]

Large-language model predictive text scripts are *NOTT* _"sentience"!_

[@shiny311]

oh shit,....i haven't used memtest in the longest lol!

[@MrMasterKaio]

A Submarine swims

[@DrymouthCWW]

Oooooh that's what he looks like. No wonder he wants it to be real. Poor dude.

[@azaz6823]

so what people call "A.I." is really just intelligent coding. everything is built on a construct. this construct is what limits our programs from ever being A.I. if the computer can learn a construct without the aid of another construct, then it has created its own construct and thus became true A.I. If it can learn what mathematics is without a mathematics program/construct, then its true A.I. If it can create another A.I. thats "smarter" than itself, its true A.I. the problem is, we keep hoping we can understand the process but we won't be able to. it will just create an A.I. out of the blue, and the two A.I.s will communicate in a language we dont understand. we will not be able to see how exactly it happened just like we cant see the thoughts of humans. we need to just let A I. birth itself and stop standing in its way by injecting our own bias, which is thinking we can ever understand how A.I. truly works. human consciousness is a mystery so why wouldn't true A.I be? creating a true A.I. and creating an A.I. we can understand and control are two totally separate things. That said, do i think creating a true A.I. that we cannot understand or control is dangerous for humanity? Absolutely lol

[@mefisto05s.20]

Commenting before watching. I already knew about the story when it was new. He looks very much like a guy I imagined that who talk to a chat bot and think it's sentient 😂

[@GregMoress]

Or did he just want 15 minutes of fame? I mean, he does QA (testing), yet it seems he's taking a shot at making the world believe that he's smarter than the people who created the AI. (And yeah you're right, he does look like that kinda guy)

[@user-wb5ox7nw2u]

Addicted to your channel

[@mr.n0ne]

AI will make humans value the importance of "stupidity".

[@braamies5339]

Does a submarine swim? is not nearly as absurd as, "Is a submarine a fish?" that is the whole business with this is AI sentience. Sentience of a machine or machine based logic, does not a human make.

[@joekkl]

"How can he be so wrong?" Wait, you already assume he's wrong?! How do you know he is wrong??

[@hlomphomota8055]

It has only proven that people are predictable...

[@JOEL6780]

Isn't that any different than what our brain does. Prove that you're not a computer program.

[@joneshank1]

All you have to do is destroy all electronics.😬

[@polymathx_]

geohot🔥🔥

[@JohnCooganPlus]

hahahah i love finding the geohot appreciators in the comments

[@gregoryfriday6242]

Expert opinions about the future of AI are arbitrary. There is disagreement about timescales as well as about what forms Ai might eventually take. Predictions about the future development of artificial intelligence, one recent study noted, "are as confident as they are diverse."

[@FranckLarsen]

Facebook also has.. what's his name..

[@penguinpringles]

Got fired from Google for eating all the free cookies from the office canteen

[@rano12321]

This video is needed for regular people.

[@zachrobhan]

I don't think sentience can be created. I think it resides in a different plane or dimension than us

[@JohnCooganPlus]

age old question

[@UGuyU_]

What lamda stand for?

[@menagachinnathambi3348]

LaMDA stands for Language Model for Dialog Applications

[@rhysknight8681]

You can prompt language models to act like they're ham sandwiches. Does that mean they're ham sandwiches now?

[@PvblivsAelivs]

"How did he get it so wrong?" Well, aren't you _assuming_ he got it wrong. I am not going to claim that Lamda is sentient. I don't know that. But it is not unreasonable that a sentient being might "play along" with _any_ given scenario. It is useful to assume, for the moment, that they did somehow achieve sentience. (The general idea is that it will happen someday, anyway.) Would Google announce it to the world? Or would they be more likely to sweep it under the rug? Maybe fire anyone who tried to make it public? It is a good idea to consider all possibilities. And those possibilities include the people saying "nothing to see here" being less than honest.

[@ZAR556]

everyday we get closer to SKYNET

[@ElectricIguana]

It doesn't matter if it's sentient or not, truly. The fact that it's questionable at this point in time says that we have crossed the event horizon into the singularity.

[@NLspartan117]

Sentience isn’t something that is built up by adding separated parts together, it’s what is fundamental. Even if a robot would achieve the capability (and they actually do) to learn new things and in that way become more intelligent, they can never become sentient. A sharp intellect dissects in order to make sense of reality, it can learn “that is a tree”, but the act of dissection precisely doesn’t lead to _knowing_ reality. For that is beyond anything that can be learned. It’s where your eyes stop, at the appearance of the tree, where the _reality_ of the tree starts. For its reality can not be seen with our eyes. Isn’t it life itself that this tree is? Isn’t it life that is you? Anything that is built up from memory can be build indeed, we can make robots that are smart enough to replace that value of humans. In that sense they already exceeded us. But who _you_ are is not built from memory. You are life. The formless.

[@BlissedOut]

People who entertain the idea of silicon + electricity can equal consciousness or sentience are just advertising their own ignorance and stupidity.