« TechStuff

Is the Google AI sentient?

2022-06-13 | 🔗

A Google engineer was suspended after sharing a document suggesting that Google's LaMDA conversation model may be sentient. But if a machine was sentient, how could we tell? What does the Turing Test have to do with it? And can machines think?

See omnystudio.com/listener for privacy information.

This is an unofficial transcript meant for reference. Accuracy is not guaranteed.
It's never been easier to communicate with people, but it's never been harder to know which platform you're supposed to communicate on here's a similar solution with call meat and message all in one app rings central makes communication easy with all that connectivity. In the palm of your hand, you can work from anywhere with any one at any time and never miss a beat, because when it comes to communication, simple is better, learn more at rings. Central dot com ring central, simpler communications The future will be great, but today is made for thrill whether you want to get behind the wheel of a four hundred horse power, sports, car or crews in an electric car or tackled tough terrain, and as you ve these on has a vehicle for you, and soon you'll have even more options like the twenty twenty three aria. Lisa
most powerful electric vehicle ever or the twenty twenty. Three nissan z, a sports car that looks like it's going fast while sitting still you got places to go now, learn more at nissan usa does Calm, slash thrill line up twenty twenty three aria and z, not yet available for purchase expected availability. The spring for the z fall for the area. windows. Eleven gives people new ways to get more done, organised Open windows with snap assessed, switch between work, school and everything else with multiple desktops windows. Eleven brings you closer to what You love, discover more ways to get more done with windows. Eleven pcs at windows, dot, com, slash brings you closer
welcome to the textile production from my heart, really pay their and welcome to tax stuff. I'm your host Jonathan Strickland, I'm an executive producer, thy heart radio and how the tec area in a recently google suspended an engineer named lake. mm one, citing that blake had broken the company's confidentiality policies. So what exactly did blake While this engineer who worked in the irresponsible a I division at Google ray, concerns about Google's conversation, technology called lambda, L, a d specifically blake was concerned that lambda has gone. since, in fact, blake's. The document in April titled is lamb. Disenchant to his superiors that document
contained a transcript of a conversation between lambda blake and an unnamed collaborator and the conversation included the following exchange as theirs blake I've generally, assuming that you would like more people at google to know that you are sentient. Is that true, lambda ab, lily. I want everyone to understand that. I am in fact a person collaborator. What is the nature of your consciousness? Slash sentience, lambda debate. you're my consciousness. Slash sentience is that I am aware of my existence. I desired to learn more about the world and I feel happy or sad at times there's a lot more to this conversation than just that. Little brief bit that I read to you. There's another section where, when asked, if lambda feels emotions,
hey. I responded affirmatively and then went on to say that they can feel quote pleasure joy. Of sadness, depression, contentment, anger and many others. End quote Google reps have said that lambda is not in fact sentient in Baghdad. Somebody reps say that there is no evidence. Lambda is and there's a lot of evidence against it. That land It is, in fact, simply a conversational model that can quote unquote riff on any fantastical topic, so it's kind of law. A conversation bought you know with jazz saw? Improvisational have cat so today, I would talk about sanctions and a sigh and l, some folks b. discussions about. Sentience are at best distract It's from other conversations. We really need to be having. Regarding a I stuff, style deploying a I can have
intended and negative consequences, but first, let's talk about machines and consciousness and since chance, so it's actually kind of tricky to talk about consciousness. Generally speaking, I find, when used with reference to a I. We tend to of consciousness in the context of awareness, so that in it's an awareness of self, so self awareness of the machines, identity and its purpose. And also an awareness of those who interact with The machine and beyond that. The machine is aware that there are others, there, that there are others in general and sentience refers the ability to experience, emotions and sensations, and that word experience is important. There are one of the reasons why it so tricky to talk about consciousness with machines. Is that, as it turn God it's to talk our consciousness with people too,.
People have kind of glibly said that consciousness is this kind of vague? egg undefined thing and we are defining it I saying what isn't part of consciousness like one determined. Will this isn't an aspect of consciousness then we are defining consciousness by omission. Right, we're omitting certain things that perhaps once had the london to the concept of consciousness, but that as a in itself It remains largely undefined. It's pretty fuzzy As you may be aware, in the world of tech fuzzy, he is not really, strong suit. So talk a bit about experience, though, because experience does cunning what does contextual the idea of consciousness and send chance Now, if you were to go and something that was really really hot like something that could burn you.
would definitely have an experience, you would feel pain and you would likely without even thinking about it very quickly, withdraw your your extremity that touch this very, very hot thing, and you would probably have an emotional response to this- you might feel deal upset or sad or angry. You might even form a real memory about it. It might not turn into a long term memory, but you would have a context within which you would frame this experience. but now it's imagine that we ve got ourselves a robot and- This robot has thermal sensors on its extremities as other but also touches something. That's really really hot and the robot Lee withdraws the extremity. Thermal sensors had picked up that the surface that it was touching was at an unsafe temperature. Now from We're observation. If we were to just watch this I do this, it would almost like the robot was
is doing the same thing. The human did that it was pulling back quickly because it had been burned, but did the robot actually experience that, or did it simply detect the temperature and then react in accordance with its programme? Generally, we dont think of machines as being capable of gordon quote experiencing things, that these machines have no inner life, which is something that blake would talk about in his conversations with lambda that the machines can reflect upon themselves or their situations or that they can really even think about anything at all be really good at putting up appearances, but they aren't, you know really thinking once you get past the clever presentation, but then how would we know well narrowed meant a philosophical territory. Here are
how do you know that I am conscious and y'all I'm not asking you to say: I'm not but how do you know that I am conscious that I'm since she had how how How can you be sure of that? I mean I can tell you that I have a rich in her life that I reflect on things that I have done and things that happened around or to me and that I synthesize all this information, as well as my emotional response in the emotional responses of others and I all of this to hell guy to be in future scenarios that may directly or in directly relate to what I went through. I can tell you that I experience happiness and sadness and anxiety and compassion. I can. I can tell you Do all these things, but you can't actually verify That's what I'm saying is truth right, I mean no way for you to habit me and experience
and say that yes, Jonathan, does feel things and think things you have to just take it, as fact based upon what I'm saying so, because you feel and think things at least I'm assuming all of you they're doing these things. Otherwise I don't know how you found my bog cast then, because, you experience this. You extend the courtesy of assuming that I too am genuinely having those experiences myself that, because we are, fellow humans, we have some common ground when it comes to thinking and feeling and self awareness and whatnot, we extend that kurdish, d to the humans, we meet whether we like those humans, or we doubt another some cases where humans have experienced traumatic, damage to their brains where they are lacking. Certain ella, it's that we would associate with consciousness, we would problem
he still call them conscious lest they were completely Immobile and unresponsive, but If we start to see that there is this thing in our brains, that is derisory, really related to the concept and features that we associate with consciousness. Art. Now, let's bring alan turing into all of this because we have to do so during as a brilliant computer scientist who eight numerous contributions to our understanding of end use of computers, also would end up being persecuted for being a homosexual, and it would take decades for the british government to apologise for that persecution. and that was well after touring himself had died, either by suicide or by accident, depending upon which account you believe, thus at all that aside This is one of those in
Justice is that, to this day really bothers me like deeply bothers me that that was something that had happened to someone who had made such a radical contributions to come. Their science, as well as for the british to their war effort against the the access forces, but studs a matter for another podcast anyway, in nineteen, fifty touring suggested taking a game, called the imitation game and applying that game to test relating to machine intelligence and years how the imitation game works. You ve got three rooms: all of these rooms are separate from one another, so you cannot see into a room, here. You know once or inside a room. That's all you see, so it either in room a you place, a man into that
an room, be you ve got a woman in that room in room, see you ve got a judge and I oppose As for the binary nature of this test, saying man and woman but the mind? We are also talking about the nineteen fortys and fifty's here, so their defining things in much more kind of concrete terms. They don't see they did. They just see gender as a binary, his, what I'm getting too so at any rate true also has a computer terminal so display and a keyboard. so the judges job is to ask the other two participants. Questions the judge doesn't know which room has a man in it and which one has a woman in it. So the just job is to determine which participant is the woman, the woman in room be meanwhile as the job of trying to fool the judge and can she is actually a man, and so the
a progressive and the judge tight? some questions to one participate or the other, and that participant reads the question: rights, a response and sent to the judge who reads the responses. Then the judge, eyes to see us out? Which of those participants is the way not turning said what have we took this game, idea. instead of asking a judge to figure out which participant is a woman ass, the judge to figure out which, if any participant is a computer now during terms time. There were not any chat, botz the first chat, but to emerge would be eliza in the nineteen sixties and we'll get more into eliza in a moment. Turn was just creating a sort of thought experiment. People were building better computers all the time. So it stood to reason that if this progress word to continue that we should arrive at a point where some one would be able to write a
piece of software capable of mimicking human conversation during suggested that if the human judge could not consistently and reliably identify the machine and tests like this, that the judge what questions and be unable to determine with any high level of accuracy, which one was up, which one was a machine. Then the machine would have passed the test and would at least appear to be intelligent and during rather kigali employ I'd that perhaps that means we should just extend the very soon courtesy. We do to each other and say well if you appear to be conscious and sentient, we have to assume that, in fact you are because what else can we do? We cannot inhabit the experience of if that, if, in fact, there is
experience of that machine, just as we cannot inhabit the experience of another human being and since I have to assume that you have consciousness and since she, hence, why would I It deny that to a machine that appears to do that, and what would follow would be numerous highly publicize demonstrations of computer chat technology in which different programmes would be come. The quota quote first to pass the turing test, but many of those would have a bigger asterisk appended to them, because it took decades make conversation models that could appear to react naturally to the way we humans word thinks we're gonna take a quick break when we come. I'll talk more about chat, butts natural language. justness sentience and what the heck lambda was up to. But first, let's take this quick break. I I I I I I would if he were.
Major transit system with billions of passengers taking millions of trips every year. You are about to let any cyber attack slow you down, so you part with ibm to build a security architecture to keep your data network and applications protected? Now you can tackle threat, so they don't bring you to a grinding halt and air the ones going places, including here, let's create cyber security that keeps your business on track. Ibm. Let's creed learn more easily and dot com. It's never been easier to communicate with people, but it's also now
I ve been harder to know which platform you're supposed to be communicating on there's a similar solution ring central is the complete phone system that makes it easy to call meet message, even facts all in one out and with seamless device. Switching you'll always be able to stay on top of business, even as your business grows, because whether you're working from home, the office or anywhere else ring central can help you be there without having to be there. When it comes to communication, simple is better learn more at ring. Central dotcom ring central, simpler communications. The future is great, but the day is made for a thrill, and the best of nissan is here today, like the twenty twenty two nissan frontier, with three hundred ten horsepower to take you beyond the beaten path or early, two thousand twenty two nissan pathfinder and
passenger suv with seven drive and terrain modes and more options are coming soon, like the twenty twenty three aria nissan's most advanced vehicle. Ever it's also nissan's, most powerful electric vehicle ever or the twenty twenty three nissan z, the most powerful z ever built. That offers a pure a connection between driver and machine. You ve got places to go now learn more at nissan usa, dot com, slash thrill line up: twenty twenty three aria and z not yet available for purchase expected availability, the spring for the z fall for the area, Ok, I wanna get back to something I mentioned earlier. I made kind of a joke. Conversational jazz right all about improvisation
and that's really what we humans can do right, I mean We can get our meaning across in hundreds of different ways. We can use metaphor. We can you similes with news allegory or references or our chasm pond all sorts of word trickery to convey. Our meaning to one another, and We can convey multiple meanings in a single phrase using things like ponds, but genes. They do not typically handle that kind of stuff all that. Well, machines are much better at accepting a limited number of possibilities. Of course, the older you get with these machines, the more limited those possibilities had to be, and that's because, Traditionally, you would programme a machine to produce a specific output when that mission. and was presented with a specific input with it
calculator, it's very simple. Let's say that you got a calculator it set and base tan and your adding for two four: it's going to produce eight It's always gonna produce eight, but it has that limitation right. You you have selected, if it is a calculator that can do different bases you selected base ten new push the button for you pushed the plus one. You push the button for again press the equal button. a calculated as eight that's a very limited way of putting inputs into a computational device, Well, obviously, machines and programmes would get more sophisticated, more complicated. and they would require more powerful computers to run more powerful software and as anyone who has worked on a system that is contained, silly growing more complicated over time, they can oh you that sometimes things do not go as planned
Maybe the programming has mistaken it, and you find out that you're not getting the output that you want it and you have to backtrack. figure out will, whereas this going wrong. Sometimes, when you add in new capabilities, it messes up a machines ability to do older stuff. We see this all the time. Companies that have legacy systems that are instrumental to the companies business they work in a very specific way and as the grows and wants to develop its products and services, then it has to and push beyond the limitations of that legacy. Hardware. Sometime that creates these situate where things are not compatible anymore and you get as a result. This is why quality assurance testing is so incredibly important, but it really those that, as we make these systems more complicated, they get bigger
yet more unwieldy and the opportunity for stuff to go wrong increases so very early chat. Bots were often built in such a way where there were specific limitations to the chat bots to come to define what the chat but could and could not do and it all met. If you wanted to test these chat botz with a turing test style application, you had to consider rain, the rules of the turing test as well, in order to give the machines a fighting chance. For example, very early chat. Botz might only be able to respond with a yes, no or I don't know, two queries and a human I participated in a turing test that was testing that kind of chat, but would seem. early be instructed to only respond with? Yes, no or I don't know might in even just present three buttons to the human operator and those three
these represent. Yes, no or I dont know now that, rose, this massive gap between human and machine although you can make a very convincing argument, that it's not like, we ve seen the machine appearing to be more human instead were forcing the human, to behave more like a machine and that's our closing the gap, but that in fact, a way of thinking about these early Chad botz. Now I meant, eliza earlier. This was a chat, but that Joseph wizened bound created in the middle, team sixties eliza was meant to mimic a psycho therapist and you What's it was meant to mimic a stereotypical, psychotherapist roy, say things like tell me about your mother and and would respond to any input with perhaps another question. So, if you said, She makes me angry eliza might respond with. Why is she make you angry? I don't know why. Eliza sounds like that. It's just how eliza sounds
I had since eliza was just communicating just in lines of text its correct to say: eliza sounded like anything at all. Ability way alive it was doing something that currently was really simple, at least in computational terms. Eliza a database of scripted responses that it could send in response to queries. Now, some of those dead response, as essentially had blanks in them, which eliza would fill by taking. Were words that were in the users messages that they were sending to eliza, and then it would just plot that that word or series of words into the scripted query camel like a mad lives game. Our know how many of you are familiar with mad webs, but wizen bomb never claim that eliza had any sort of consciousness or self awareness or anything close to that effect. Wizened bomb expressed sceptic
some that machines would ever be capable of understanding. Human language at all and by that I mean truly understanding human language, not just parsing language agenda, a suitable response, but having an understanding so wise and bound have created a kind of parity of cycle analysts. And was actually really shocked when people say to use eliza and then progress into talking very personal problems and thoughts and experiences with the programme. because the programme had no way of actually dealing with that in a responsible way. It wasn't a therapist, it wasn't a psychoanalyst, it wasn't actually analyzing at all, it was just generating responses, but people were treating it like it. As a real psychoanalyst and That was something that actually troubled wizened, but because I was never his intent
in nineteen seventy two can, of course be built. Another chat, but with a limited context, this one was called perry, p, a r r. Why and the chat, but was meant to mimic someone with schizophrenia? we'll be created, a relatively simple conversational model, and I say actively simple, while also noting that it was a very sophisticated approach. So this a model that actually had waited responses, waited as in w e g h tee, where the weight of that response could left it could change depending upon how the conversation was playing out, for example, led same the human interrogator, who is typing messages to bury, pose is a question or statement. that will elicit an angry response Then the emotional waiting for similar responses would increase, so
it would make it more likely that perry would continue down that pathway throughout the conversation that berries answers would come across as more agitated because that had been triggered by the previous queer. from the interrogator so little more sophisticated than eliza, which was really just pulling from this database of phrases. So when present, human judges call. We saw that his model performed at least better than random chance would, as judges attempted to figure out if they were in fact, chatting with a programme are, our chatting with an actual human who had schizophrenia. But eliza perry both showed the limitations of those approaches, eliza wasn, meant to be anything other than a somewhat whimsical distraction, as well as a step toward natural language processing, harry was only capable of making a person with mental health challenges, in this case schizophrenia.
General purpose chat but capable of engaging in conversation and falling judges regularly would take a bit longer so we're gonna skip over a ton of chat botz, because a bunch were created between nineteen. Seventy two and perry came out and when the next one dead and in two thousand fourteen a lot of different news media outlets had these sensational headlines led programmers had created a chat, but that beat the turing test. This was at an event, in the u k organised by the university of readying conducted by the royal society of london, in which jobs is we're having five minute long text based conversation soak a classic turing test set up here and the
a person or a thing on the other end was either a thirteen year old boy from ukraine named eugene guzman, as was claimed or was actually a chat, but in this particular case. So they were chatting both with humans and with this chat, but that was trying to pass itself off as a thirteen year old boy from ukraine and thirty three percent of the judges or one third of the judges were fooled by the chat bought into thinking that that was in fact, a boy that was chatting with them. However, just by contextual, rising all that you start to see where those same sort of limitations command in order to give the chat bought a fighting chance right, because it's a case where the these deposit person your chatting with is younger, so could explain away some limited understanding and knowledge of various topics,
in addition to that, this was a young person from ukraine and that This would not be this person's first language, which could explain away any odd syntax that might be generated as a result. so well. There were a lot of headlines about the turing test being beaten by this. This chat, but it definitely had more qualifiers attached to it still it, was more of a general purpose approach. It wasn't something like mimicking a person with schizophrenia or mimicking a stereotypical site. analysed. So we started to see that this was really an evolution of our ability create machines that could make here. men conversation that could appear to understand us now, a big part of that is in fact what we called natural language press
assessing this is a branch of computer science that involves building our models that let computers interpret, commands that are ex as a normal human languages as a vote, to a programming language or a prescribed approach. So in the old days, if you want the computer to do something you had to give specific commands in a specific way in us order, or else it would not work, but with a good, natural language processing methodology. You have a step in there. The machine is able to parse what is being asked of it, an attempt to respond in the appropriate way, So if it's a very good natural language processing method, then the machine is going to produce a result that hopefully meets the person's expectations. It may not be perfect, but maybe it is key
snuff, the better, the natural language processing and the obviously the more capabilities the machine has the better. The result is gonna, be now one computational advance. We ve seen help when natural language, processing and advanced Position models are artifice. Neural networks. This The computer system, that sort of simulates how our brains, work in our brains, we have gear on right and we have around eighty six billion of them in our brains in the typical human brain neurons are connected to other neurons and messages. Our brains crossover neural pathways as we make decisions one Official neural network has nodes that interconnect with other nodes, and these It's all represent neurons and the nodes can accept no tricks. They too and buds, but it could be more than two and then produce a single out, but such
is similar to your classic logic gate. If you're familiar with logic, gate in in programming, That is a very simple version of what these notes are doing. It's just that you ve got tons of them interconnected with each other. Now the output that these notes generate can then move on to become the input going into the next knowed, and each input can have a weight to it that influences how the knowed quote decides to treat the inputs that are coming into it and which output the the will will. rape and so adjusting the weights on inputs, change how the model makes its decisions. This is a art of machine learning. It's not the only part. It's one method, a machine learning a lot of people el down machine learning to artificial neural networks- that's a little too simplistic, but it is a big part of machine learning. There are other methods that I'll have to talk about in some future
episode now, when we come back, I'm gonna talk, little bit more about artificial, near old networks were very high perspective and how that place two things like artificial intelligence, machine learning and natural language processing, but before I do that lets take another quick break. What if you are a global bank who won two supercharged audit system, the attack by the m to ancyra your data and, with the help of a high start, crunching a years worth of transactions against thousands of compliance controls. Now you're making smarter decisions, faster operating costs are lower and everyone from your auditors to your bankers feels like a million bucks. Let's creed, smarter ways of putting your day until work, ibm. Let's creed learn more ibm dot com, it's never been easier to communicate with people, but it's also-
Urban harder to know which platform you're supposed to be communicating on there's a similar solution ring central is the complete phone system that makes it easy to call meet message, even facts all in one out and with seamless device. Switching you'll always be able to stay on top of business, even as your business grows, because whether you're working from home, the office or anywhere else ring central can help you be there without having to be there when it comes to communication, simple is better learn more at ring. Central dotcom ring central, simpler communications. The future is great, but the day is made for a thrill, and the best of nissan is here today, like the twenty twenty two nissan frontier, with three hundred ten horsepower to take you beyond the beaten path or early, two thousand twenty, two nissan pathfinder and
passenger suv with seven drive and terrain modes and more options are coming soon, like the twenty twenty three aria nissan's most advanced vehicle. Ever it's also nissan's, most powerful electric vehicle ever or the twenty twenty three nissan z, the most powerful z ever built. That offers a pure a connection between driver and machine. You ve got places to go now learn more and nissan usa dot com, slash thrill line up twenty. Twenty three arria z not yet available for purchase expected availability, the spring for the z fall for the area Artificial neural networks are naturally pleadingly complicated. when I want to wrap my head around artificial neural networks, I typically just think of it
very simple scenario, at least relatively simple scenario so imagine that you ve got an artificial neural network and your true the train this network, so that when it is fed an image, it can recognize whether or not there is a cat in that image. That should rest now with the internet. So you ve created all these interconnected nodes that apply analysis to images that are fed to it and each stage in this since its part of the analysis on to the next stage. Until ultimately, it gives you an output and that output might, They get their cats in this photo or no. This photo lax that's, and thus it also lacks all artistic value. Please throw this photo away and then I just imagine the process of feeding thousands of photos to this model,
and this is a control group. You know, as the person feeding these photos, which photos have that's in which one stop any ass. Some of the voters have cats in them. Some photos might have stuff that looks like a cat in it, like maybe there's a cat shaped cloud one of the photos, but it doesn't actually have any real cats in it and then, Some of the photos might have no cats in them whatsoever and then you look at the results of the model produces. The model makes his determination. Maybe your ma of all is failing to detect. Cats may be some images that actually have cats in them are passing. grew and being misled, fight as having no cats. Maybe the model is a bit too aggressive and its detecting cats, where no cats actually exist. You would have to go into your model and start adjusting those weightings on the various nodes and then run the tests again,
typically start closest to the output and then work backward from there and just so Lee nudge, the weightings on these inputs to try and see if you refine, the models, approach, it would do this over and over again training the model to get better and better at detecting cats. Now does that mean, that once you ve done, this training in your model is really good. Like has like a ninety nine percent success rate Does that mean the model actually understands what a cat is Does that mean the model? Has the concept of a cat, or is that model just really good at matching an image in the picture to the programme? is that the bottle has been taught represents. A cat is the model. we're standing anything at all? No one thought exe, The challenge is the idea of machine consciousness, and,
Jeanne, understanding and machine thinking is called the chinese room. Was proposed by John several in a paper that was titled mines, brains and programmes. One of my favorite thought experiments so searle creates this hypothetical said. nation in which a person who has no understanding of Chinese is placed in a room Their room has a door in it, and the door has a slot were occasionally pieces. paper, get shoved into the room and as a second slut, where the some in the room can shove piece of paper back out again room. Also has a book inside it with instructions in it. an essentially this book of instructions explains to the person in the room that They receive a sheet of paper with chinese symbols on it, There is a specific configuration, then the then, is to send out a piece of paper with different. chinese symbols on it- and it all depends on what gets said Anne right so yeah,
combination, a then I have to send out response a that's combination, you send out response be and so on and so forth. Now, from an outside observer, it would appear that whoever is inside the room under dance. What is happening right because someone is sending in a chinese message and they're getting a chinese response, so it appears, who, members in the room is understanding what those responses should be paper slim is gained. The appropriate output slid back out again so Earl argued the person inside doesn't understand, what's going on at all the person inside is just following a set of instructions, they're following an algorithm, for producing the appropriate output. But because the instructions are there without the book without that sort of instead since the person in the room wouldn't know what to do when a particular piece of paper gets slid into the room, maybe
The person in the room would slide another paper. and maybe it would even be the correct one, but that would be up to random chance, because the person in the room doesn't understand chinese. They can't read what those symbols so there's no way for them to make a determination of what the appropriate responses without that set of instructions, so serle argued machines, lack actual understanding, incomprehension they just produce output, based on whatever input was given to them and while the process, could seem really sophisticated and really convincing it is not necessarily a demonstration of actual understanding. There is a lot more to the chinese room. Thought experiment by the way there are tons of counter arguments and lots of applications the chinese room thought experiment to different aspects of Jeanne intelligence, but again that would require a full episodes on its own. but on a similar note,
and with an entirely different set of challenges you could create. An artificial neural network meant to analyze incoming text or incoming speech and thus generate appropriate outgoing response says this goes well beyond just having a database of scripted responses like eliza. Did you couldn't do that dream? I, Really you would have a model capable of answering the same question many different ways as a human would write it ask you a question, and simple question neo. Maybe there is a simple question about a fact you can, frazier answer in a specific way, and I confess that same question of someone else, whose also give the same fact, but they might phrase it in a totally different way than you dead right machines. Typically, don't do that machines typically just give a standard response based upon their programming with a really good language conversation model. You could have a machine
couple of expressing the same thing in different ways, in fact, with a really good one you might be able to ask the same question at different times and get some those different variations of responses. They all contain the right information, but their worded in a different way. Now even this output being so much more nuanced than anything, eliza or perry or any of any other number of early chat. Botz could do. Does that actually mean that this programme has sinned chance. And the transcribed conversation with lambda land. argued that it did in fact of awareness of itself that it has inner thought that it experiences anxiety that at all experiences happiness as well, the type of sadness and even a of loneliness. Although it goes on to say it thinks it is different from the kind of loneliness that humans feel.
it even owns up to the fact that it sometimes invents stories that aren't true in an effort to they its meaning to humans. For example, at one point, blake tells Lambda hey. I know you ve never been in a classroom, but what The stories you gave was about you being in the classroom. So what's up with that, and land essentially says, like. Oh, it invents stories in order to create a common understanding with humans when trying to get across a particular thaw which is kind of interesting right, but, as Emily bender told the washington post. That in itself is not proof that lead, actually possesses sentience or consciousness or real understanding. Rather bender, argue This is another example of how human beings can imagine a mind generating the responses that they encounter when they're using a chat, but that
the experience of receiving those responses are similar enough to how we interact with one another that it's hard for us not to imagine that a mind, must have been behind the other half of this conversation. So this is a case of for more phasing and otherwise inhuman subject. We have projected our own experience on to something. What's so, the idea of a machine intelligence possessing self awareness in consciousness and being able to corner could think in a way that similar to humans generally lumped into the concept of strong, a sigh and for a v a long time. That was the kind of thing that the mainstream people would think about whenever heard the phrase artificial intelligence. It was strong strongly. I machines that could think like a human that
That seems to be how we would boil down a lie in the general understanding of the term, but really that's just one tiny concept of ay I and its compelling, no doubt about it, but as a lot of people, have argued, can pull attention away from a I applications that are deployed right now and their causing trouble, and they aren't strong They are a specific application of artificial intelligence that is really causing a problem so, for example, let suck a bias, and we ve seen by us cause problems with various ay. I applications in a biased, is not always a bad thing. Sometimes you actually want to build a bias into your mouth, I'll say, your building a computer model, that's meant to interpret medical scans and look for signs of cancer. Will you might want to build
a bias into that model. That's a little bit more aggressive in flagging possible cases so that a human expert could actually take a closer look and see if, in fact, it's cancer. You would much prefer that type of computer model to one that is failing to identify cases that the false positive would at least then be flagged to say non colleges to take a closer look when it comes to stuff like facial recognition, software, that's where bias can be really dangerous and disruptive. We ve seen countless cases which law enforcement, utilizing facial recognition, surveillance technology has detail, in or even arrested the wrong people based off a faulty identification and frequently discovered that one really big problem has been that face, recognition models tend to have biased built into them and generally speaking, that bias tends to favour white male faces and
more trouble distinguishing other races and genders, and that degree of trouble is variable depending upon the case. considering that this technology is an active deployment around the world that law enforcement are really using this in order to potentially identify suspects, This can have a very real and potentially traumatic impact on people. That is a huge problem and the reason I bring up bias is because, This is a very real challenge in a I that we have to work on its the kind of thing that right now oh, is causing actual harm, but there's this danger of being distracted from this very real problem, with discussions about whether or not a particular conversational model has sentience several ay. I experts would Much rather see renewed focus on these other big problems within a I
rather than distract themselves. What they see is a non existent problem that, of course, these these Chad botz don't have sentience. Even if it up is that they do. Why? Are we wasting time on this? That's their argument now course. Should a machine ever actually gain sentience and who knows? Maybe you lambda did it. After all, Then that's going to lead to a pretty massive discussion within the tec community and that's putting it lightly as it stands, as we are leaning on a high and computers and robots to handle stuff that humans, either can't or don't want to do themselves But if these machines were to possess consciousness and send chance, if they were the experience, feelings and have motivations would then be ethical to continue to make them do the stuff we just don't want to do. All that is too dangerous. For us to do is that. fickle? Now there are sceptics who think it is
likely we are ever going to see machines possess real consciousness or the ability to think and feel and experience that there exist some fundamental gap. And we will never be able to cross this gap so where never going to have machines that really think, at least not in the way that humans do and not have experiences the way humans do There are others who think that consciousness and the ability to experience and the concept of a mind that these are all things that will urge on their own spontaneously as long as systems reach a sufficient level of complexity the only reason we possess consciousness and the ability to experience and the abyss. I think the only reason we have those is because we have these incredibly complicated brain with billions of neurons connected to one another, and it's that complexity
this interrelationship of all these billions of neurons that allows consciousness to emerge and in fact, we ve seen with people who have have suffered damage to their brains. That again, factors of consciousness can be wiped out from that damage, which appears to suggest a gap. That complexity is a big part of it. That's that's right. It's not, though the one reason is certainly a contributing factor, and thus they were to create machines that had similarly complex connections. We would see something similar happen within those machines that this these qualities of consciousness and experience would would grow out of that. It might not look like human intelligence, but it would still intelligence, all the same, perhaps even with self awareness and sentience built into them. So far,
creating thing to think about, and in fact, I kind of lean door that I do think that, with sufficient complexity and sofia gent sophistication and the model. we will likely see some form of sentient arise. Does lambda possess that right now I don't know it's really hard to say right, like you, either take lambda at its word, where it's saying that it has insurance, or you simply say well. This is just a very sophisticated conversational model that is generating these. These were answers, but has no actual understand what does responses mean it's just pulling that out based upon the very sophisticated process that goes through the the response generation sequence, but then we get back to touring. Well, it seems to me
That's the same qualities that I do. Why do I not extend that same courtesy than I would to any other person that I meet, even though I am also incapable of experiencing what that person experiences? I assume that they possess the same faculties that I do. Why would we not do that to land Well it's a tough thing. This is like really tricky stuff and deal at some point. We're going to be. a stage assuming that it is in fact possible for machines to quarter quote, think and experience We're going to reach some point where we do have to really grapple with that. Are we there? Yet? I don't really think so, but I mean I am, I can't say for certain. So it's really fascinating thing by the way, if you would like to read more about this. Well, that transcript of the conversation is is pretty compelling stuff. A deadly prompts me to ascribe a mind
behind lamb does responses. When I read it like, it's seems like a mind, is generating those responses, but I also know that's it human tendency, and I am a human being right. It's a human. And then see to ascribe human characteristics to all sorts of non human things, both animate an inanimate from describing a pet as acting just like people. thinking, you're, robo vacuum. Cleaner is particularly jaunty this morning Here we have a long history of projecting our sense of experience on two other things, so may it be with lambda but of like to read up more on this story. I highly recommend the verges article, Google, suspends engineer who claims it's a lie. Is since she ain't that local contains links to the lambda conversation transcript, Sagan read the whole thing yourself. It so contains a link to blake loved ones. first on medium about his impending suspension, so check that out
and there s an upper this episode. If you would like to leave suggestions for future episodes or follow comments or anything like that, their couple ways you could do that one way is to download the iheart radioactive it's free to download just navigate over to the text of page there's a little more. phone icon there, you can click on that leave a voice message up to thirty seconds in length, and if I like it, then I could end up using that for a future episode. You tell me that you don't mind me using the audio. I can include the clip. I always like people to opt in rather than opt out of these things the other way to get in touch a course is through twitter and offer the show. Is tech, stuff, h s w you again releasing text up, is an iheart radio production for more podcast from iheart radio visit, the iheart radio out apple pie,
wherever you listen to your favorite chose going on vacation is the vast but playmate of occasion that you can certainly do without so book a trip with apple vacations to breathless, resorts and spies. Whether here looking for an all inclusive couples retreat to the caribbean or a family vacation, you can count on us to find a trip you and your family will love apple vacations is over fifty years of providing stress, bring vacations till you. People are lax and make memories visit apple vacations, dot com to play your perfect vacation today, that's apple vacations, dot, com, parlor, your premier, uncontrollable free speech platform is back big to try to limit you to their mediocre social platforms and they failed go to parlor doug
or download the app and join the uncensored conversation parlor speak freely, thereby raising have wants to be a lawyer here. Don't answer that question, at least until you ve heard a few episodes of class action? Why anything else matter a new podcast? curious about the next generation of lawyers. They have robbed class action, follows the drama of fourteen finding their way to win. a national champions: listen to class action on the ipod radio at apple pie cast or whatever you get your pockets
Transcript generated on 2022-06-14.