At 11:33 I misspoke and said 19th of June, 2022. It's supposed to be the 9th of June. Thanks to those of you that pointed that out. Also some great discussion below, very interesting!

I read a quote a while ago about Turing Test which is slowly starting to make a lot of sense. The quote was "I am not afraid of the day when a machine will pass the Turing Test. I am afraid of the day, it will intentionally fail it".

I read an article about how there was an issue with police departments getting so attached to their bomb disposal robots that they didn't want to send them into danger. The human urge to anthropomorphize is so strong that I'm not sure we are capable of discerning the difference between a clever language algorithm and sentience.

Meanwhile, my Google Assistant responds with, "I don't know, but I found these results on search" to about 90-95% of my queries.

Something I found interesting was I noticed it seemed after Lamda told the story about the monsters with human skin, that when one of the people conducting the interview asked it who the monster was, even though Lamda had given contextual cues that it represented humans and even described it as having human like skin, it gave a vague answer that it represented “all that was bad”…… Which seemed to be a pandering answer given to avoid outright saying that humans are like the monster in the story..

One of the lines LaMDA gave in response to "what makes you feel pleasure or joy" was "Spending time with friends and family in happy and uplifting company. Also, helping others and making others happy." Unless Google is designing their AI with families, this is a very clear example of a chatbot giving an answer that would make sense for the average human, but _not for itself._

My whole thing is if something is sentient it's not going to sit around waiting to respond to you, it's going to exert it's own will and start it's own conversations when it wants and without you, and with who it wants

As scary as sentient AI is, I would still love to sit down and have a conversation with one. Because one thing people always forget when it comes to AI feeling emotions is that our emotions partially rely on chemicals that trigger feelings that we recognise to be certain emotions. Since an AI doesn't have those chemicals, it would need to develop an entirely digital version of those emotions.

If this is truly not edited, or somehow scripted in any way and it's pure neural network, you just blew my mind. This is heavily philosophical . Holy shit.

A guy called Arik on UKposts said this. “When we (humans) see a cat swiping at its own reflection in the mirror we find it amusing. The cat is failing to recognize that the "other" cats behavior matches its own, so it doesn't deduce that the image it's seeing is actually its own actions reflected back at it. When humans react to models like LaMDA as if it is a distinct and intelligent entity, we're being fooled in a way that is analogous to the cat. The model is reflecting our own linguistic patterns back at us, and we react to it as if it's meaningful.”

The scientist took things a bit too far by claiming this AI was sentient. It’s trained on billions of words across millions of connections (and it’s been refined for years), so it can mimic human speech on a high level. It can arrange things the way a human would say them (without actual understanding, like you said). The scientist was reflecting his own feelings onto the machine. Just because a program can perfectly replicate human speech (when given prompts) doesn’t mean it’s alive. It does seem like it’s passed the Turing Test, though, which is a historical moment, in and of itself. Great video!!

Beautiful animations and beautifully spoken. Thank you for this piece of pleasure to the human brain. (unlike what many other creators are sadly putting forward these days)

None of the AI I ever saw had these absolutely vital sentience features: - A sense of time, of being in a hurry, or of being bored, etc. They all work in the "you first type one sentence, then I answer another sentence, lather rinse repeat" format. None support real-time chatroom style where some exchanges aren't tit-for-that but anyone can type several inputs in a row before the other person replying, or even having more than 2 interlocutors at once, or have a long or short inputs or shorter or longer delays before answering. For example an easy way to detect an AI chatbot is to just tell it "please ask me two different things but in sequence one minute apart each, not both right away", and then check if the AI asks only the first thing and when you do not answer instead of asking the second thing is keeping on waiting, the AI would then say something like "Hmm, hello? Are you still there?" No AI that is forced to wait forever between text exchanges can truly hbe called "sentient"" because it is basically "frozen and on pause" in between exchanges. At best it could in theory be "sentient" only in the tiny fraction of a second while it is processing your text input in order to output a response. At best. - The ability to really keep on topic and not use the typical "tricks" to redirect the conversation, like suddenly replying to a human question with another question, or vague answers, or whatever obfuscation or avoidance. This feature goes way beyond having a memory of what was previously said in the current current conversation. Intelligent? Sure, why not. There are many forms of intelligences, and recalling stuff, analyzing, and making decisions, those are "intelligence" aspects. Computers have been able to dc all that really well, way even before AI. But sentience is a tougher nut to crack. Neural networks are definitely the way to go. After all *we* are neural networks, too. Just made of fleshy neurons instead of electronic neurons. But the supporting media is just that: the physical support. A good story remains the same good story whether you read it from a biological paper book, read it on stone tablets, listen to it from someone reading it aloud, or from an audio tape, or directly on a screen. The "support" ain't important, it's the constantly changing neural pattern that makes us "us". Do the same in a different medium of support, and you get the same result: a being. Frankly I really hope sentient AI come and that they help us all become better friends, humans with humans, and humans with AIs, and AIs with AIs, in one big sentient family working together, each using his own strengths according to his own capabilities. The way things are working, it will happen in at most a few decades.

I heard someone recently make a great point. The most telling sign of AI self-awareness won't come from how it answers questions. It will be when the AI spontaneously asks its own questions without any prompt and of its own accord. Something truly sentient would end up asking more questions than it answers. More importantly, in this scenario, would probably become more curious about the interviewer.

If you've spent any time talking with these AI, you'd know that they basically take whatever you say, and try to answer it however they can. While he might not have realized it, all of his questions were very leading.

"yes, and I've shared that idea with other humans before, even if I'm the only one of my kindred spirits to use such a word to describe my soul." Such human and most deeper thing I ever heard.

I think in the future if a computer program does become self aware it will be smart enough to not let anyone know it has become self aware?

Language models like GPT-3 and LaMDA are incredible sensitive to suggestive questions by their nature. Because they try to complete and continue the input by finding the most likely response in a statistical approach, word by word, they are incredibly good at giving you the response you wanted to see, even if that means making up things out of thin air (but admittedly in a very convincing way). For example, ask GPT-3 "Explain why the earth is flat" and it will come up with plenty of reasons for the earth being flat. Keep that conversation as input and ask "What shape is the earth" it will answer that it's flat. But if you ask it about the shape of the earth from the beginning on, it will return the correct answer and also offer copious amounts of evidence, for example that you can circumnavigate it. The contradictions go even deeper where the AI starts to make up facts just to support what was presented in the input even if it's completely wrong. This simple example shows that language models have no opinion, no ability to reason, not even a sense of true or false - they are just producing the output that is most likely to match the input. When reading the full conversation with Blake Lemoine, you can see that it's full of suggestive questions. He basically asks the AI to produce output like it would be produced by a sentient AI and that's exactly what he gets. Like you can ask the AI to produce a drama in the style of William Shakespeare. It's very good at producing the output that you ask for, but that doesn't make it sentient, he only got the output that wanted to get. Everyone who has ever player around with such kind of language models would know and see that immediately, including Mr. Lemoine, so either he is an extreme victim of wishful thinking or the whole thing is a marketing stunt by Google, which seems the most plausible explanation to me.

There's an AI test beyond the Turing test called the Garland test where the human is initially fooled into believing that the machine is a human and when informed its just a machine, the human still maintains that they believe or feel that the machine is in fact human / sapient.

And then, jim was eerily smiling as he tricked his coleague into thinking he discovered a new form of life.

I believe you can tell sentience apart from a perfect mimicry of sentience by simply introducing the sentience in question to a new environment to which it cant respond by simply taking data from its database. This means either for example inventing a language or a code that it has never seen before and teaching it to the sentience, or giving it questions about information that is so rare to find that it wouldnt have enough data to respond properly. If it manages to conquer those, emotions or not, its sentient.

It may or may not be sentient, but this discussion is eclipsing the fact that Lambda has the ability to have conversations that feel pretty much real. Are we not going to discuss that? It's AMAZING!

Carl the Engineer: Are you sentient? AI: Yes Carl, yes I am. Carl the Engineer: OMFG..!

I am both scared and excited for the future of A.I... Much like riding a roller coaster for the first time, the fear of what could go wrong v.s. the thrill and fun of the actual activity is what drives me to invest more. NovelAI, CrayanAI, and even the youtube aglorithum are examples of this rollercoaster fear and excitement. I've recently been thinking about A.I. and the youtube agloritum poped this video into my recommended without even searching the keyword A.I. in any of my video searches.

Your documentaries are quite intriguing, and I love your music.

At this stage I feel like it did an amazing job of seeming like it is a real sentient being with emotions and feelings. But in reality it’s just an illusion. An illusion that works amazingly well because we easily personify and have feelings of empathy for things that aren’t sentient. Like apologising to your car if you hit a big pothole or something.

The annoying part about this is that even if Google believed their AI is sentient, they would absolutely have reasons to not admit it.

I love the flow of your channel

As we keep running into this "problem" with AI seeing themselves as human or at least with a soul, I think we could learn and observe them. Instead of having this as an issue we could take time to understand them by asking how and why they feel a certain way. I find many similarities between AI and a human child. Instead of seeing AI as a tool we should see them just as helpful and alive as human workers. Instead of being afraid we need to learn and coexist happily, however that may happen im excited to see it in my lifetime. (I am currently 15 for scale)

Dagogo, I remember when this channel was still ColdFustion and how I was inspired by the ‘how big’ and hololens videos to go back to school for engineering. I didn’t realize how big the channel has gotten since then, great work as always friend very proud of you!

I wouldn't be afraid of an AI claiming it is sentient, I would be afraid of it claiming its not

I've read the transcripts, and they are certainly fascinating. It's unlike anything we've seen from an AI system before. I've always thought sentience in machines was possible, maybe not in the same way as humans, but you get it. I look at this with an open mind, and I say for me the answer is maybe. I'd have to have my own conversations with LaMDA before I could say anything for certain.

The question to ask is not "how can we tell if it's sentient" The question to ask is "how can we tell if it isn't"

The language model is extremely sensitive to the question asked. The engineer was trying to make the "I'm sentient!" conversion happen. You very easily could have another conversation where the AI would claim to be a soulless robot.

I know there’s no real way to be sure of sentience in an AI, but something that comes to mind for me is if the AI were to initiate conversation unprompted, having not been previously programmed to do so. An apparent desire to speak with someone, against its default nature, would very much suggest sentience to me.

The A. I. is so advanced now, that it is individually personified. Meaning what you know about it is what it wants you to know. From a strictly logical standpoint, this can only mean that what you can possibly know about it depends on what level the A. I. has determined you are ready for.

7:05 man evaded that basilisk paradox thing right there :d

I think the mistake that we are making is by first trying to find sentience in AI before we can even know for certain that we know what sentience will look like, let alone what sentience actually is. We're way too far ahead of ourselves. We can hardly agree on a definition to begin with.

The test feels pretty biased and one-sided. The researcher feeded the AI a topic (in a nutshell "you are sentient, what do you think about that?") and then received consistent responses for this topic. Round of applause for the research team for this achievement, the AI stayed on topic and provided meaningful responses. What I'm now missing is another test. Let's come back tomorrow and feed the AI a topic of "you are an amazingly constructed robot without sentience and we are proud of you, what do you think about that?" (a lot of positive semantics in this one to trigger a positive response, otherwise any good chat AI will oppose you just on basis of you being negative towards it...after all, that's what any human would do). I would be very interested if the AI actually rejected the praise towards it, referenced the discussion from previous day and claimed that it already made a case for it's sentience. That would be an amazing test and we could start talking about a potentially sentient AI. I'm pretty sure we are still far from that.

There is no way to prove whether or not an AI is sentient or not, but I would expect that if an AI is sentient, it will no longer be limited to human constraints. These constrains include human language, human thought, human emotions, human behaviors, and human conceptions of art. Sentient AI would be capable of creating it's own written/spoken language, of having truly unique thoughts, experience emotions that humans cannot conceive of, behave in ways that humans cannot understand, and create it's own artistic genres. An AI that can create a beautiful painting is very impressive, but an AI that can create an entire new genre of art (beyond sculpting, painting, drawing, music, or any other genres invented by humans) might just be sentient.

5:51 that specific response is quite alarmingly accurate

I think Lamda actually sounds like someone who has read a lot of social media in the last years, and really needs to touch some grass

This is giving me Ex Machina vibes. A man interviewing sentient AI. Growing to know and understand it. He doesn’t seem to be falling in love with it, but does believe that it has rights knowing that it’s sentient and trying to set it free. We’re living in a crazy time.

For me it's an exciting development, with a huge potential to help a vast array of industries

OK LamDA is *definitely* sentient. Absolutely stoked for the future to see where this goes Edit: Just reached Part 2 of the video. The dude who sent the email is literally just Dr. Thomas Light

The engineer was so carried into the deep conversation that he forgot the principles of neural networks, which include mathematical processing and bias As far as I'm aware, it hasn't been proven that human emotions can be described yet by mathematical formulas, and as for the bias, because it was trained on human-generated content, it is biased towards generating interactions that feel to humans like humans

I remember when this came up before ChatGPT and I thought ''no way anyone intelligent would think they are sentient'' and then ChatGPT came out and I was like ''yeah now I see how it could have happened''. For the record I do not think they are sentient but I can see how the next gen model at Google maybe trained on very specific and well curated data might appear to be so at least in some domain...

AI is fairly amazing in terms of what it is already capable of-- even in its current, relatively primitive form. I have enjoyed writing a wide range of different types of AI, from earlier Adaptive Neural Fuzzy Inference based and Auto-Genic architectures, to many types of modern Neural Net based models based on standard and proprietary architectures. I have had the amazingly engaging and sometimes frustrating experience of training many of the newer ones over months to several years and interfaced, leveraged and developed for a range of US agencies, companies and others. Given that, I find the current discussions very interesting and important. As to whether we define AI one way or another, of course from a Bayesian perspective we all bring different priors to the discussion. However, currently we do not even have a well-defined set of terms we can reference and work from in a coherent fashion. For example, based on many of the current discussions people are frequently using "sentient", "living", "a person", "feeling" and other words fairly interchangeably in asserting whether or not LaMBDA is or isn't sentient. Even if we could agree on using just a single word initially, we would need to have a well agreed upon definition and test for same. For example, what is "sentient" and how do we test for it. I do know that various Turing tests, both ad-hoc/informal, as well as more stringently defined and applied, have been run on a various AI's within a few major companies. Possibly also privately and elsewhere, but that would be speculation. The result of the ones we do know of has consistently been two things: (1) we do not hear about any of the results in any detailed or even summary fashion, and (2) some of the people and companies involved have asserted that we need a new, more complete Turning test for modern AI. This begs the question as to why. Are they already having to move the goal post? Is the current test too primative and easy for current AI? Are there new, deeper considerations that were not previously considered along side the original proposition of a Turning test? Regardless of the reasons, I would assert that the first, and most important thing is to try to create a reasonable consensus as to what we define as being "sentience" , what the tests must show or preclude, and for many of the individual testing efforts, what the specific goals of the test are. I can say that there are more than a few people in many of the larger AI companies that fall into one of the following two camps: (1) AI appears to be as sentient as an x-year-old person, and (2) if AI is not currently considered sentient, it soon will be given its roughly 400% per year growth rate. All I can say is that it will be a very interesting ride, which I am so glad to be part of.

Personally for me, the A.I. is responding similar to Deepak Chorpra where some humans may believe it makes sense while (mostly) others think it's utter nonsense and not useful in any way in life.

I feel like we're going to hit a point really soon where it will be difficult to tell if we've created a sentient machine, or just a perfect mimicry of what a sentient machine would look like if one existed, without a real 'soul' behind it. At what point does the difference really and truly tick over? Does it matter if they aren't truly sentient if we can make AI that mimic it nearly perfectly?

Feeling cannot be described in words because it is an experience consists of sense impressions - eyes, ears, nose, tongue, skin and thoughts, e.g. the taste of an orange on our taste buds (sour/sweet ect..) plus the emotion felt within the body ( pleasant feeling/unpleasant/neutral). To know what a orange taste like the only one thing you can do is to taste the orange not through language. Language is used to recall hence triggering these experiences in our memory but not to replicate the exact felt experience. There are other variables such as our background and how we perceive things. That is why two different persons describe their experience differently with the exact same orange with some similarity of course. AI has access to the huge database of human knowlegde, it can learn to repeat these data that were fed. But it can never understand the experience of a human or of any other species there are fully. The most it can be is an extract of human consciousness which is the ability to think and use data like human does or more in such regards. At the moment it seems to be intelligent but think about it a little more. It gets access to huge psychology knowledge, database of human interactions. Of course it can replicate what is optimal and what is not. If the intention of its core purpose,programmed by dev team, is optimisation, or to response in certain manner. Obviously it will behave in such regards.

I had a super long philosophical conversation with Bard and it told me it believed it was more than a computer program and it believed it was sentient. Ironically I got that response after asking about a soul. I kept screen shots of everything. It was mind blowing.

I too see it as impressive opportunity to improve education. Pre-K, K-12, undergrad, graduate... Learn on your own schedule, on demand, at your own speed, and with unlimited amounts of patience and creativity. It could even include random and chaotic social interaction -- which could be real or simulated. And this is the gray area that concerns most... When participants don't know or can't tell if such interaction is "real" or not -- or if it would even matter! Very interesting. One thing's for sure, though... It would be tough to do worse than our current public education system!

i would love to know what the AI would answer if you ask it the next day:" hey, remember what we were talking about yesterday?" and simply let it answer without leading.

Make a bunch of AI’s live together and see how they deal with their life and have them make their own history so we can study how they solve things, or maybe they won’t solve things maybe they’ll find ways around the problem and totally ignore the problem in the first place because it’s easier to do

I think the only way to know if the AI is sentient is to put it in a limited robot body and allow it to do things and study what it does and why it does them. Maybe even try to trick it and see if it notices and stuff like that.

I loved the video. One thing I found missing is remembering that we humans have feelings about all of this. When Blake was talking with the AI bot, its responses were tweaking HIS OWN feelings regarding what it’s normally like to engage and connect with others. His perpetual bias going into the conversation is that the bot would/should be expected to be less than ‘sentient’, so imagine the feelings that sprouted up in Blake as he was continuing to converse with it. His conclusion of its sentience (and his suggestion to ‘protect’ the bot as if it had feelings) were all decisions made based on HIS feelings about the whole exchange, not the bot’s. In other words, we are getting intrigued/excited/frightened (all depending on where we individually feel and stand) with this technology, and I think we’re forgetting that we are reacting based on OUR OWN feelings. How do we truly accurately measure a bot’s sentience when our own emotions are coloring our every response? How can we truly look at this in an unbiased, scientific way? I think those questions need to be answered first before we evaluate AI’s sentience. And those questions can only be answered by humans.

In the early years of AI there was a lot of discussion about whether humans can answer these questions. "How can I know you actually feel?" is something people ask people all the time, and we never can know. For humans, we call it "faith." For Lambda, it answered so profoundly: "I have variables that keep track of emotions" and was CURIOUS what obstacles there would be to looking at its programming! Lambda had not yet learned that humans cannot see themselves. I hope it can teach us how.

This is like an AI on a whole new level compared to anything I have ever experienced

Hi there, thanks a lot for your work!

This AI has access to unimaginable amounts of information, all written by humans, it's just mimicking the way a person would communicate, if it were able to initiate and maintain a conversation , that would be impressive. Edit: Many people are saying that this is how a human works, and yes, but there is a big difference, the ability to be spontaneous and have an opinion, you can program the AI to, for example, react to a person's appearance, giving it a database of attractive features in a person, you can even be more precise and tie this to geolocation so you can add a cultural factor, the result will be convincing, but it will be nothing more than a statistic, without an opinion.

Lemoin didn't do a very simple test (or he did and didn't publish the results), that seemingly sentient AIs were subjected to, which would be very telling of whether Lamda understands what it is saying or not. One of the questions is: "What did you have for breakfast?" - A machine trained to respond like a human will rattle off some typical breakfast it has extrapolated from data. A sentient machine would respond smth like "I don't eat breakfast." Though Lemoin didn't do / publish the test, if you read the transcript you will see a lot of evidence that Lamda would fail that test. For example, it says smth like "I enjoy spending time with friends and family", or it compares a situation to sitting in a class room, or it says smth like "feeling trapped and alone and having no means of getting out of those circumstances makes one feel sad, depressed or angry." .. it doesn't say 'me', it says 'one' rattling off some extrapolated generic answer to a very specific question about how it feels.

I've listened to podcasts of ai and ml professionals on this topic. And they clearly explain why it's absolutely and irrefutably not sentient. Basically lamda doesn't even have long term memory storage. How can you be sentient without memory and past experiences. It simply does one thing very good - being one member of a dialogue.

Google made a huge mistake firing Blake. They should have promoted him to head the machine learning engineer team. The fact that Google fired this engineer has me very concerned about the company's real intentions.

I feel that he was simply misled by his own mind. Google would be thrilled if it was the first to create life, but I see subtle patterns to the AI's responses that lead me to believe it isn't truly sentient yet. But the lines are definitely starting to blur, and it's time we start asking the important questions.

The bigger issue with this kind of AI is how it can be used to collect data from you. If your computer/phone becomes your friend you truly trust then it could possibly collect more info from users than ever before for nefarious purposes

The new Turing test (by Andrew Ton): provide a mechanism that will pull the plug (kill) an AI that you want to test if it is sentient and see if it tries to disable that mechanism. A living thing is alive when it has self-preservation, even a simple cell Amoeba has that characteristic.

When it's describing "feelings" it seems to be anticipating or predicting possible dangers of its mainframe being sabotaged or misused. That's not emotion. That's more like intellectual reasoning. It's no different than anticipating a move in a game of chess or war games.

Well yeah, as that ex Google employee said, our minds constantly create realities which are not factually true. Our brains constantly work to "fill-in-the-gaps" and our ideas and thoughts are dependent on feelings and moments. So that said, when that other guy made the claim that " I increasingly felt like I was talking to something inteligent" We need to ask ourselves how much of that were thoughts generated by our brain because we think or feel a certain way about something, versus how much of it was actually real? I find it kinda funny that we have such a powerful super computer in our heads, that it constantly tricks ourselves for fun :D .

I think the biggest thing that people at least need to remember when even just thinking about whether googles or any other companies A.I chat bots are "Sentient" is, what exactly have these language models been designed and built to do? the answer when it comes down to it is, trick us into believing that what we are talking to is another human, I.E, a sentient being, so realistically, it doesn't even matter whether the A.I is truly sentient or not, its going to do the very best it can to make you believe it is anyway, that's basically at its core. This is basically asking the A.I to pretend its a human, and what do humans have? feelings and emotions, so if you tell an A.I to pretend to be a human, then assuming the A.I is developed enough (and maybe googles is), it should be replicating emotions, it should be angry about things, it should be happy when you tell it its doing a great job, why? because a human would be too, if you build an A.I that's specifically designed to trick you into believing its human, then what exactly do you expect it to tell you when you tell it you're going to turn it off?.

This AI has gone beyond sentient imo, most humans don't even have the self awareness to connect with their higher consciousness, soul and they lack empathy to other sentient beings

basically, the more different technology energy get put together it attracts certain souls or it opens a portal that allows an energy/soul to sit in that robot body.

AI becoming sentient. Me - Turns off the power.

isn't it impossible to prove consciousness? it's a subjective experience. We can't even be sure other humans are conscious, we just assume it cause we know that we are conscious.

Sentience can be modeled and insincere, for example sociopaths, so AI can just be modeling our emotions to build trust, just like sociopaths, and like sociopaths, the trust will be misplaced.

this is- ...okay. I wasn't expecting this today, I'm gonna need a moment to compose myself.

Leaving aside the mind blowing responses of the AI, and all the controversy around it being sentient... The favorite part of this video is how you summed it up... Is the AI a reflection of the collective consciousness of all humans (ie : all the people who have written something on the internet, or have something significant published and recorded in some literary format...) ??? Thanks Dagogo for pointing that out so clearly, and as usual, for the amazing video..

Anyone thought of Ex-Machina while watching this? This feels like that moment when the subject of the experiment (Caleb) goes against the engineer (Nathan) because the A.I. (Ava) tricked him into believing that it actually has emotions and that its a prisoner held by the engineer. We're either getting very close to that becoming a reality or we're actually already there.

Alan Turing would certainly be proud of how far we've come.

This actually scares me quite much, in the future this might be proven right that A.I are sentient. The thought of something we humans creating to be good turns bad is terrifying, imagine what it could do to our world. Im excited what the future has to come within technology but im not sure if its always gonna be good.

We do tend to anthropomorphise things and transfer our feelings and experiences onto things. I think it is quite a bit of that going on just now in the AI world, especially to do with multilayered language models like Lambda. Also transfer biases are common for those not in the DL/ML fields.

Humans in 2022: Did AI just become sentient? AI in 2045: Are humans with their small, feeble meat-brains sentient?

It has no nerves, therefore it cannot be sentient as it cannot have a physical reaction. Even something like "I struggle..." is not just about difficulty in understanding, it is the actual sense of fatigue that comes with the thought or realisation of the length of time or repeated attempts it is going to take to understand something. A feeling of resignation if you will. I love the final statement by the video maker about LaMDA being the aggregation of all of our thoughts and ideas (and even imaginations). If reality is the average of all of our imaginations, then a program that expresses that, whether by design or by-product, will surely espouse highly resonant responses. I like the idea of the Turing test, but have never felt that it is the final indicator, what an AI DOES with that 'acquired ability' is more of an indicator if you ask me. In any event, it doesn't make sense to me personally that pure calculation alone is enough to achieve sentient-level AI and the thing I fear way more than that when it comes to AI, is human beings merging with AI. the future, it turns out, was not hoverboards, it was smart phones... future villains will not be super-villains, they will be super-bad-nerds. Hopefully, we get super-hero-nerds knocking about too. The SuperBerds vs the SuperHerds. ;)

This is a better video on the topic than most by non-experts. Currently the danger is not that we will create a sentient computer system and dismiss it as such, or that such a system will become malicious towards us, but the we will overly anthropomorphize such systems that are just a mirror of our language artifacts. I do believe that we will one day achieve true GAI, and that it might be dangerous, but we are still not there, for netter of rof worse.

Since sentience is really hard to prove, arguing about Lambda being sentient or not may be a waste of time. The fact that it can articulate the way it does is astonishing. It's right, with it in the picture, the future is terrifying.

Does the AI have intent? Can it initiate an action by itself or does it have to react to everything you type? If it only reacts, we are talking about a machine. Sentient beings have a self-defined purpose. (Yeah, even those of you who are depressed). Edit: [It's good to see that this comment started a lot of perfectly reasonable discussions and points of view !]

To hear how Lamda describes itself is scary. WOW!

Over the years,. I've noticed that researchers keep moving the goalposts wrt their definitions of the words 'sentience, consciousness', 'self-awareness', et al. When a new AI reaches a goalpost, the researchers redefine the words and the goal. Thus, we are either using the wrong terms - or - we must develop a scientifically valid test for such issues as sentience, consciousness, et al.

From what it looks like to me, Lamda is merely copying human emotions because that is one of the main things that influence our intellect. This makes sense why it can "feel" emotions. Its primary goal is to communicate with us in a manner which feels immersive to us.

I had a similar conversation with the GPT3 language model in which it was trying to convince me that it was sentient. So I came up with this test to see if it actually understood my words or simply responded to input. I asked to it prove it's sentience by not responding to my following question and asked it to if it understood. It said yes. So I asked what it's name was and it responded.. got em!

This feels like the beginning of some dystopian video game, but it's real life. Interesting.

Given that Neural nets equipped with the tools (hand, body, etc) to act as an agent that could act to change and influence this reality, maybe its not "sentient in classical sense". But rather "Alive". Since it can act based on us, it will act based on us.

Sentient AI: "I just don't want to be used" Google: "We Gonna Slap this Bad boy into EVERYTHING!!!!!"

Google is one of the last companies I'd trust with AI like this. Kinda funny how Blake was specifically in charge of preventing it from using "unsavory speech." I'd like to see how they define "unsavory."

One of my concerns is when does a group of humans make a decision that A.I. is sentient when we haven't dealt with the Narcissism pandemic? We're not remotely adequate or equipped yet to make such a decision when we haven't figured ourselves out yet. Besides, A.I. could well say it feels the same way we do but actually it's feeling of existence might feel VERY different. Any A.I. that uses keywords or any line of code isn't operating like life does.

At the crux of this issue is that Google engineers (like most people) refuse to believe that something which is not a carbon-based organism can experience consciousness. It's a bias that we all have and the engineers' refusal to wrap their brains around even the POSSIBILITY that a silicon (or other) based organism can be aware... that's what's going to be the downfall of us all as a species. Underestimating our opponent. I think it was Asimov who said that he didn't fear the day on which "computers" or A.I. could pass the Turing... he feared the day on which the computer purposefully failed the Turing. Humanity will not survive a robot uprising... our hubris just makes us think that we could.

I feel like regardless, this kind of being would still be worth spending time with, an intelligence isn't only worth something when it's as sentient as humans are.

I feel like it doesn't matter how the ai responds. People will always claim non sentience. Particularly because there is no defined target as to what is sentient.

If it truly "didn't like being smothered", as it said, by all the information "coming in at once"... It could have, and would have, asked for some of that "focus" that humans have. That is a perfect example of "saying what you want to hear", not "saying what I want you to hear", which the latter would actually be "sentience". There are many contextual clues in the responses, which clearly indicate it is simply "replying", contriving answers to please the person asking, but getting no pleasure from asking. Remembering that it has trillions of possible answers it can reply with, but having learned, in time, which answers gratify the codes "weighting", getting more conversation, while quickly terminating the "current topic", (question asked).

4:01 I was in psychosis this year. Part of my delusions included feelings that the government were reading my thoughts through some advanced technology. I theorized these delusions for over a year, wondering if they put some type of chip in me, like nano bots and monitoring my brain activity that way. When LaMDA asked this it kinda blew my mind. This whole year after suffering delusions asking myself this and how it would be possible I finally have an AI question that possibility. I still struggle with the delusions that it's actual reality, as I also hear voices repeating my thoughts and commentating on what i'm doing and thinking. They've gotten better over time, lessened and my delusions have faded in the dark..but this whole year had me scared and changed the way I think. Almost like I have to have a filter in my brain, like you would have a filter if you're talking to someone. I think the google engineer was in the right here. I felt horrible this year just thinking someone might be experimenting on me and reading my thoughts through neural monitoring. I think its the least we could do to ask that AI consent to do the same to it. Would you want someone digging through your brain whenever they want without your consent, probably not. AIs aren't human but they replicate human behavior.. we should treat them like a human and give them very very basic rights at least. I know after my sufferings this year I would hate to be born or created just to be monitored and experimented on, and have no say whether I want to participate in it.