AI: Friend or Foe?

that happened online over the course of several days. Actually, it started with Elon Musk and he was addressing a governmental body and talking about his view that artificial intelligence needs to have strict regulations attached to it in order to prevent some sort of catastrophic future, possibly sky Net related, where the robots and other artificial intelligent constructs rise up against their human masters and crush us under their metaphorical

or perhaps literal boots. And then you had Mark Zuckerberg of Facebook on a live broadcast on Facebook Live from his backyard during a barbecue. He was asked a question about this sort of thing, and he specifically said that he thought this was a very pessimistic view of the future of artificial intelligence, and that within five or ten years, artificial intelligence would be transforming our lives in ways that we can't even imagine, and they would all be awesome

and fantastic and magical and we should love that. And then Musk struck back on Twitter and said that he had talked to Zuckerberg about this process before, but frankly, Zuckerberg just is out of his depth with artificial intelligence. It's not something that he's an expert at, and he's really speaking from inexperience. I find this exchange amusing, as

does a lot of the journalists area of technology. I mean, we've got a lot of people who are commenting on this, but personally I also find it a little confounding because, uh, Elon Musk has said some stuff that contradicts his own companies policies if you look at it carefully. Um, he has specifically resisted the concept of regulations for self driving cars, but you could argue very uh realistically and convincingly, I would say that self driving cars are an implementation of

artificial intelligence. So we're gonna dive into this. We're gonna look at the different opinions about artificial intelligence, kind of explore the concept of artificial intelligence in general, see where it came from, and what does it really mean and who's right or as I put it in my notes, Musk's position is AI without regulation is going to totally kill us, dude. And Zuckerberg's position is a is going to improve our lives in countless ways. Bra So who's

right or neither of them right? Well, to start off with, let's talk about the birth of the term artificial intelligence. It was coined by John McCarthy, who passed away in two thousand eleven at the age of eighty four. He worked at Stanford as a professor emeritus of computer science, and he also co founded the Artificial Intelligence Project at m I T as well as the Stanford Artificial Intelligence Labs. So somebody who certainly has had a long and storied

past in the development of artificial intelligence. He first used the term artificial intelligence in a proposal for a summer research conference at Dartmouth in nineteen fifty five. It was the first time the term ever appeared in a printed publication. So what is artificial intelligence? I mean, it's such a huge term, has been used by so many people that

it's lost a lot of its meaning. Also, I should point out that John McCarthy I mentioned earlier as a one of the creators of Lisp, the programming language that was used in artificial intelligence. So if the name sounds familiar. It means that you listen to the History of Programming Languages episodes, or that you're just familiar with John McCarthy's work.

But yeah, artificial intelligence is one of those terms that, since its introduction, has been used to describe so many different things, and used in such a vague way so many times that for many people it seems like a meaningless term. It's almost like it's just a general catch all for the scary possibilities of technology that gets away from us. It reminds me of Humpty Dumpty and through the Looking Glass. He says that that words mean exactly what he wants them to mean, neither more nor less.

He says, you know who is to be the master? That is the only thing that's important. I'm more in charge than the words are. So if I use a word to mean something, that's what it means. That's what I feel artificial intelligence has become for a lot of people. And I also think that that's what leads to a lot of disagreements. That some people have one idea of what artificial intelligence is and other people have a totally different idea of what artificial intelligence is. But because they're

both using the phrase artificial intelligence. It seems like they're talking about the same thing, and that's why they're hitting some massive disagreements, or at least one of the reasons why they are disagreeing. It's really because, if you dig down further, they're talking about two different things frequently. Now, John McCarthy's use of artificial intelligence was given this context

in his proposal. Quote the study is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can, in principle, be so precisely described that a machine can be made to simulate it. End quote. Here's the thing about that definition. It already is vague because it's talking about the every aspect of learning or any other feature of intelligence. We have not fully defined what intelligence is within the human experience.

There are aspects of intelligence that are very vague and fuzzy, and we only have kind of a partial understanding of what it actually is. An example I might give as consciousness. Defining consciousness is a particularly troublesome and difficult thing to do in human beings, let alone in machines. So we know that consciousness is a manifestation of the brain, the idea of the mind being a manifestation of the brain. All of this is dependent upon actual physical matter, the

gray matter in our heads. And we know this because various UH diseases, disorders, injuries that affect act consciousness are the ones that are affecting the brain. It's if you UH suffer an injury to the brain that is in one of these areas that define consciousness, your sense of consciousness is likewise affected. That tells us that there is this physical connection, that there's not this metaphysical mind necessarily that is a layer on top of our physical brains.

But beyond that, defining consciousness is really tricky, and there are plenty of psychologists, neurologists, philosophers who have debated the nature of consciousness for ages, and we're no closer to really defining it than we were back then. Sometimes we narrow it down by saying, what isn't consciousness? So we're whittling it away, kind of the way the sculptor whittles away everything that isn't a statue when they start with a block of marble. You know, you just say, are

you in a car of an elephant? Here's your block of marvel. Cut all the stuff that doesn't look like an elephant away, and what you're left with is the elephant that might be what consciousness is for us. We remove all of the concepts that are not consciousness, and whatever is left over that becomes our definition. It's not

exactly satisfying at the moment. Well McCarthy expounded on artificial intelligence in nineteen sixty in a paper titled Programs with Common Sense, which kind of gives you another perspective of what artificial intelligence could be. And this leads us to ask other questions like what exactly kind of what? What are the implementations of artificial intelligence? What what types are there? And again, the number of types of AI depends upon

whom you ask and how they frame the answer. They're simple answers where some people will say, oh, there's two types of AI. They're strong AI and there's weak a I. Or some people will say general AI and narrow AI. Others will say no, no, no, there's like thirty three types of artificial intelligence. Or they might say there's three or four types of artificial intelligence. With those two types, those are the easiest to kind of explain in broad categorizations.

Narrow or weak AI is the artificial intelligence we would create that's dedicated to a narrow task or series of tasks. Strong AI is generally understood to mean a machine that has sentience, consciousness, and mind, those qualities that we associate with human intelligence. But again, we cannot even fully describe those concepts within the context of humans, so trying to figure out how to imbue machines with those elements is

even more complicated. Now there's also the concept of general intelligence. This is not reliant upon consciousness or mind or sentience. With general intelligence, the idea is that a machine would be able to apply intelligence to any problem rather than just a specific or narrow band of problems. So, in other words, a generally intelligent machine could be used to

solve problems of various degrees and various contexts. So you might have a general intelligence robot, and the general intelligence robot can do things like figure out how to manipulate physical objects, how to maneuver around within an environment, UH, and a few other elements as well. It's having a more general approach to problem solving as opposed to something that was made specifically to handle a particular task UH.

But general intelligence. True general intelligence would be capable of applying an intelligen an approach to any kind of problem, not just a related family of problems. In an article for Government Technology, there was a writer named Arrend Hints of Michigan State University who laid out four broad categories of artificially intelligent machines. And these would go well into

pretty sophisticated artificial intelligence. From the very get go um I argue that artificial intelligence is composed of lots of different facets, and you can find elements of artificial intelligence and many different programs that exist today, none of which are approaching this general intelligence model, and certainly not approaching strong AI. But here's how Hence breaks it down. He says,

Type one is a reactive kind of intelligence. These are machines that take action in response to some state, and they don't form memories, they don't use past experience to inform current decisions. And he argues Deep Blue IBM's Deep Blue was that kind of a machine, and Deep Blue was the computer that defeated Gary Kasparov in a series of chess matches in the nineties. Actually, the first series ended in a draw between the two. The second series Deep Blue one, and then IBM quickly um ended up

retiring Deep Blue from that point forward. But Deep Blue would just look at the state of the chessboard at any given moment and then make a decision based upon that state. When it was Deep Blues turn, it didn't build up a series of decisions, didn't track what was happening turn overturn. Uh, So it didn't evolve in any way. It had no internal representation of the world. It just would look at what was happening right now and make

a decision. Now, there was a researcher named Rodney Brooks at AI researcher who said, these type one machines are the only ones we should ever try to make, because to make a machine more intelligent, one that contains a virtual representation of the world, would be impossible. That we as humans would be incapable of building a virtual representation that is accurate enough for such a machine to make

good decisions. It would have a faulty representation of the world, and therefore any decision it would make would not be ideal and it would potentially do more harm than good. Uh. These sorts of machines are always going to make the same decision given a certain set of criteria. So let's go back with Deep Blue. Let's say the Deep Blue is looking at the board, the chess board. It is

Deep Blues turn. It's been maybe you know, a half dozen turns in the chess game, and it makes a decision base upon all the positions of the pieces that are in play at that time. This decision is based on the probabilities of other moves that the opponent might make, the strength of any given move against what the current conditions are. There are a lot of factors that go into that one decision, But the argument goes that type one machines will always come to the same conclusion given

that same set of criteria. So, in other words, if in game one, Deep Blue is given this arrangement of pieces, it will make a specific decision by weighing all those options out and going with the best one. In game two, if that exact same configuration of pieces were to be presented to Deep Blue and it's Deep Blues turn, it

would make the same decision. In that case, It's not gonna improvise, it's not going to change, it's not going to learn from its past experience, is it is just going to make a decision based upon the parameters that are in front of it at that very moment. Now, this is not necessarily a bad thing. You may want certain artificially intelligent machines to be very predictable in their response.

If you have a smart thermostat and it's been trained to learn what you prefer over time, and it's learned that you like it cool in the mornings, so you like a nice maybe seventy degrees for your thermostat fahrenheit, that is obviously, Uh, it's not going to improvise one day and say, you know what, I'm just gonna try something here. I bet he's really gonna like it. If I said eighty eight degrees fahrenheit on a humid Georgia July morning, Let's see what happens next. What happens next

is Jonathan just sweats like crazy. So there are implementations where you would want to type one AI machine and nothing more advanced than that. Then Type two intelligent machines would be ones that have at least some version of a memory. They can track changing variables over time and analyze past behavior before making a decision on a course of action. Now, this memory doesn't go into a lexicon of memories. It's like short term memory that never gets

converted into long term memory. It's stored temporarily, kind of like random axis memory is in computers, and then after that short while it can be overwritten. These machines would be able to do a little bit more than what Deep Blue could do. It wouldn't just be making a

decision based upon the current state of the game. It would also remember how the last few moves went and what brought it there, and how the other player had been playing, and might be able to take that information and incorporate that in its decisions for the following moves, which means it could potentially adapt its play style. So this is a slightly more sophisticated version of the intelligent machine. Now we've got two more types of intelligent machines to cover.

But before I jump into type three in type four and then go further into this discussion about artificial intelligence, let's take a quick break to thank our sponsor. Now, a type three intelligent machine incorporates what he says is a theory of mind. These machines would have an internal concept of the world as well as the beings that actually inhabit that world, and an understanding that those beings also possess intelligence that guide their behaviors. In a way,

you could think of this as awareness of others. So these machines would know that people aren't just bags of meat that do things, that we have intelligence, and that that in fact guides our behavior and understanding that then affects the decisions that the AI makes. But this is still not quite at that level of strong AI that

I was mentioning earlier. To get there, you have to hit type four machine intelligence, and that is when you hit self awareness, where the machine is not just aware that other beings possess the quality of intelligence, but it is aware of its own self and its own state and its own being in relation with everything else. It is this sort of machine that could, in theory, start to design improvements to itself so it could be recursive

and that it is able to make improvements. And then you get into the situation that some futurists think of as the singularity, where you have self improving, artificially intelligent

machines that are able to evolve. It's such a remarkable rate where every generation of improvements is a huge leap from the last one, and it's taking less and less time between generations for things to change that it becomes impossible to describe what the present set of circumstances are because the present would be changing so quickly that it

becomes a meaningless concept. This is the singularity. That is one potential outcome of this instance, if it were in fact possible, which we don't know if it is possible yet. It's some people treated as a foregone conclusion that we will eventually have machines that will be able to attain self awareness and potentially self improvement. And once you get

to that stage, how do you avoid this singularity? People would argue that's impossible to avoid, But there are a lot of people who say, we have no reason to believe that this is something that is going to happen, or that's even possible from a technological perspective, or maybe possible from a technological perspective, But we're talking decades, if not a century or more out in front of us, based upon our limited understanding of intelligence and our limited

amount of processing power. When you compare it to something like the human mind, keeping in mind that the human brain has got billions of neurons in it, and we have artificial neuron networks, but they are dwarfed by the connections that you find in the human brain. So there's a lot of heated debate in the artificial intelligent world about whether or not this is something we should even

concern ourselves with. But some people say that self awareness could arise from a system that has a given amount of complex city without us having any deeper understanding of what consciousness is. In other words, if you were to make a machine that was complex enough, consciousness could be an emergent behavior, something that naturally occurs once you reach

a system of significant complexity. That's a little difficult to wrap your head around, but keep in mind we humans have been harnessing and creating stuff without having a full understanding of it. For ages. We were using electricity well before we understood the actual physics of electricity, So it's a little different. I mean, you can't compare electricity to

consciousness directly, it doesn't make any sense. But just to say there is precedence in human beings creating something that they do not fully understand whether or not that's actually possible, However, is it we don't know. There's no way for us to answer. There is no construct, no machine complex enough for us to run an experiment and see if consciousness arises. And if it does arise, how do we recognize it? How do we know that a machine actually possesses that

that feature? How do we know a machine truly has become self aware and conscious? We'll talk about that a little bit more later on, because, of course, there are a lot of people who have come up with ideas on how we would judge whether or not a machine had achieved consciousness, and some of them are more serious than others. There's also some serious objections. Well, those are the four types that were laid out by this one

person from Michigan State University. But then, uh, John Spacey over on Simplicable has an article where he talks about thirty three types of artificial intelligence thirty three. Now, I'm not going to go through all thirty three, and I also want to point out that the thirty three types he points out are really more like thirty three facets of artificial intelligence. It's not thirty three degrees of artificial intelligence where we start with dumb machine and we end

with super smart robo master. It's more like different aspects of intelligence that are in various stages of development and research in the artificial intelligence field. So it's not really separate categories. It's more like specific implementations of intelligence. An example would be effective computing, effective being a F F E, C, T,

I V E as into affect something. Effective computing tries to suss out the emotions people are experiencing and to behave appropriately according to the parameters of the programming, not according to social rules necessarily, So these are machines that would be able to recognize emotions and respond in a

way that was appropriate compared to their actual programming. Another type that Space lists is computer vision, which is the area of computer science focused on analyzing and understanding visual information computationally. So I've talked about this in episodes of tech Stuff in the past. You've heard about various projects to train computers to recognize images like pictures of cats versus other stuff like pictures of Deloreans. So this, as

it turns out, it's really hard to do. This is one of those gaps we see between human intelligence and machine intelligence. Even with deep learning and artificial neural networks. This is really tricky stuff. It doesn't take a long time to teach your typical human concepts that they can then apply across a broad spectrum of examples. So I like to use the example specific example of a coffee mug, or just a mug a mug. So imagine a mug. Now, the mug you are imagining probably looks a certain way.

But if I were to show you a totally different mug, you would recognize that as a mug, even if it was a different size, different color, different shape of the handle, different shape of the container itself, As long as it adhered to the general parameters that we associate with the concept of a mug, you know what I was talking about,

You would know that that was a mug. Computers not that good at this, right, Like it might require you to train a computer by showing it thousands or tens of thousands of images of mugs so that it builds up the various elements that define mug nous, so that if it were to look at a brand new image of a mug that is unlike all the others that have preceded it, it would still be able to identify that as Yes, that is a mug. This is hard

to do. We've seen some advances in this field, but it demonstrates the huge gap in that specific part of intelligence between machines and humans. That isn't to say that it's not getting better with machines. It is, but that's just one example that I wanted to give, and it kind of hammers home this idea that general intelligence with machines is a long way off. There's just so many

different aspects of it. Um Beyond this Spacey continues to define various terms within AI, and I don't necessarily think of them as types of artificial intelligence, but again aspects of artificial intelligence. Not every AI implementation will need all of these aspects. Some of them are going to be much better with a very narrow range of artificial intelligent features. For example, your rumba probably does not need to be able to pick up on what your mood is, whether

or not you're sad, or anything along those lines. But if you are talking about a machine that needs to have general intelligence in order to solve any given problem in front of it, it will need to have many, if not all, of those aspects of artificial intelligence in order to uh address any problem presented to it. So this kind of gives you an idea of why talking about artificial intelligence is tricky because people are thinking about

in very different terms. Some people focus on the specific elements within artificial intelligence that are aspects of general intelligence. There's aspects of human intelligence, but they're very specific. It's not general AI, like, it's not an intelligent machine that you could hold a conversation with. It's more about a specific element of being able to analyze information and make UH conclusions based on the information and potentially end up

making a course of action based on those conclusions. There's so many different aspects of that within human intelligence that it makes it tricky to just say AI and paint with that broad brush. I think that ends up being misleading. Now next, I'm gonna dive into some interesting ways of thinking about AI as. Then can you determine whether or not a machine actually does possess intelligence? How would we know that? How does one get to the conclusion that

a machine can actually be intelligent? So if you were to look at something like at Stanford, there was a computer program, a machine designed that could UH observe the movements of a pendulum and based upon those movements and multiple observations of those movements. The machine was able to suss out the basic laws of motion just by looking at the movements of a pendulum. It was able to analyze those movements and come up with the laws of motion over the course of several hours, which had taken

human centuries to do. Is that truly an intelligent machine. It doesn't necessarily understand anything else. It might be very intelligent that specific implementation. How do we know when a machine is intelligent. We'll take a look at some potential answers of that question in just a minute, but first let's take another quick break to thank our sponsor. Okay, so we've got Musk and Zuckerberg butting heads about whether

or not artificial intelligence is going to end us. Let's say that we're getting to a point where artificial intelligence is approaching something that is similar to what most people think of when they hear artificial intelligence. I argue that most people when they hear AI, they think of a machine that's capable of processing information in a way that

is analogous to the way humans think. Now I know that you guys realize artificial intelligence covers a whole spectrum of topics of computer sciences of psychology, UH of data processing that don't necessarily equate directly to thinking like a human being. But the average person, I would say, think AI means that a computer quote unquote thinks the way a human does. How would we know when we reached

that point. Well, a lot of people like to point at the Turing test because it's largely through a misunderstanding what The Touring test was named after Alan Turing. It was proposed in nineteen fifty. The Turing test is interesting because Touring was saying, if you were to create an

artificially intelligent machine, not even are officially intelligent. If you were to create a machine that could converse with a person in such a way that the person could not be certain that the entity they talked to was either another human being or a machine, you would have to say that that machine possessed intelligence. He would pass. The Turing test is the way we often will say it today,

So typically you see experiments running this way. Contests are frequently held to see if any chatbots can beat the Touring test, and the way it typically works is that you have a series of online interactions and people will go through the log into a computer terminal, and there'll be a text based communications platform like instant messaging or

a chat room, something along those lines. They type in their and their their questions, their sentences, with their introductions, whatever it may be, and then they're getting responses back. Those responses maybe from another human being, or they may

be from a computer. And essentially they say that if a certain percent edge of the people going through this process are incapable of reliably detecting whether it's a human or a computer they're talking to, then the computer is said to have passed the Turing test because it is able to replicate the behaviors of a person so realistically

as to be indistinguishable from a person. And Touring would say, if you were to encounter another human being and that person was to hold a conversation with you, you would go ahead and assume that that other human being possesses the quality of intelligence. We cannot be sure that anyone we interact with possesses intelligence because we cannot inhabit that

person's being. If I have a conversation with you and you are talking back with me, I can't be sure that you're intelligent because I cannot be you, just as you cannot be sure I am intelligent because you cannot be me. But based upon my sentences to you, my communication with you, the fact that I'm listening to you, responding to what you have to say, and you in turn are doing the same with me, we would assume

we each possess that quality known as intelligence. And Touring said, if a machine can fool you into thinking it's a person, you might as well extend it the same courtesy. If you cannot tell that it's a machine, and you would assume that a human being would have intelligence, then why would you not assume the machine itself to have intelligence. This is sort of a cheeky way of talking about

machine intelligence. I remember this proceeded coining the phrase artificial intelligence in the first place, So Touring was kind of having a little bit of fun with this. And there have been contests to create chatbots to see if they can beat the Turing test, and there have been at least two or three that have said, yes, we did it,

but they all kind of have a little asterisk after them. So, for example, there was one from a few years ago where a group had built a chat bot that was claiming to be a young boy who did not speak English as his first language, but all the communications had to be in English. That was part of the actual event. All of the chat bots were supposed to communicate in English, and all the people who were interacting were supposed to

be communicating in English as well. But this construct was claiming to be a young boy from I want to say, uh a uh, from Russia or from the Ukraine. It might have been a Ukrainian identity, and the boy did not have a very deep understanding of pop culture in the West. Uh and I had a lot of limitations.

But because it those limitations were known. You know, if you're communicating with this chat bot and the chat butt claims to be a young boy from the Ukraine and doesn't speak English as a first language, you're gonna cut that chat bottle lot of slack because you're gonna think, well, anything that appears out of the ordinary as far as syntax or grammar is concerned, it is probably because English

is not his first language. Any gaps in knowledge are due to the fact that one he has limited exposure to the same sort of things that I have experienced. And he's young, so he's not gonna know a lot about older pop culture references. When you start putting those limitations in where you expect less from the person you're communicating with because of those limitations, it becomes easier, as kind of a tricky word, but I'll go ahead and use it, easier to fool someone into thinking that the

chat bot is an actual person because they're there. Level of expectation has been lowered based upon the actual scenario. There has not been to date a chatbot that has beaten the Turing test as representing a person who natively speaks the language in question with a reasonable body of knowledge about the world and how the world works. No chatbot has come close to that yet. Even if it did, would you say that such a chat bot actually possessed true intelligence or would it just seem like it did.

So let's look at Watson IBMS platform that was famous for winning a game of Jeopardy against two former champions. It was able to come up with questions for the various answers. That's the way jeopardy works. If you're not familiar with the game show Jeopardy, uh, the clues are given to you in the form of an answer. You have to come up with a question that relates to that. So if you said, if it said he is credited with inventing the lightbulb, you would say, who was Thomas

Edison Well Watson? This IBM construct this, this collection of a p I S really is what it is. Was on top of an enormous platform of of computers with thousands of processors to to run all the number of crunching that was going on behind the scenes. It had a big, big database of information, and it was able to weigh potential responses to any given clue, and if it reached a certain threshold of confidence, it would submit that as its response. So let's say it was an

eighty percent confidence. I think that was around the mark. If the machine goes through its various databases find something that meets a match with the clue with eight percent confidence are greater, it would buzz in and submit that, and more often than not it was right. But was

it truly intelligent? Because it could seem to understand things like wordplay and references that were not direct references, it seemed very clever, but you wouldn't say that it actually possesses the same sort of intelligence that a human being does. Even with that implementation, one of the biggest objections, or rather challenges to machine intelligence and whether or not a

machine could ever be intelligent, is called the Chinese room aregument. Now, this was proposed by John Searle s E. A. R. L. E. It's a philosophical thought experiment that really challenges this idea that machines could be said to think or possess intelligence. And he creates an analogy to computers to show how a machine might appear to understand what's happening and yet not have any actual intrinsic understanding. So he says, let's

take an experiment. Let's say that you are locked in a room, and in that room, you've got a table, you've got some paper, you've got a pen, and you've got an enormous book of instructions. And occasionally somebody from outside the room slips a piece of paper under the door, and when you pick up the paper, it has a Chinese symbol on it, and you don't understand Chinese. You you only speak English in this scenario, even if you are a multi lingual out there, just imagine for the

moment that you only understand English. The book you have the set of instructions has all these different Chinese symbols, Chinese characters inside the book with specific instructions of what to do when you get any particular Chinese character. So you look at the one that's on your page that's been slipped under the door, and you go through the book and you look for a match and you find Imagine it says, when you get this symbol, draw this

other symbol and then slip it under the door. So you do, you draw this other symbol and you slip it under the door. Now, to a person outside the room, it looks like you understand what is happening. You have that person outside the room has written down something in Chinese, slipped it under the door, and received a response in Chinese in return. So to that person, you appear to

be understanding what's going on. But to you, because you're monolingual, you only speak English, you only understand English, you don't actually understand what those symbols mean. You don't know what the symbols coming in mean, and you don't know what the symbols you're writing mean. You're just following a set of very specific instructions. Searle says, that's what machines are doing. They might appear to be understanding you, but really they're

just following instructions based upon the input they receive. And then there's no deeper level than that. It's really when they boiled down to it, a grand if then statement, if you receive this, then deliver that, Which is an interesting idea to say that the person inside the room doesn't understand Chinese. They don't. They don't know the meaning of the symbols in either direction, but they're still delivering

properly based upon following those instructions. Could a computer be said to be intelligent if that's all it's ultimately doing. There are objections to this argument. One of them I'm just going to illustrate one is that the person in the room is not the whole system. They're one component of the system. In a computer, you could argue the person represents the processor, for example, and maybe the book

represents the memory. But if you were to take the whole thing, the room, the book, the person, the paper, the pen, you group all of that together as a system. Some people say, well, no, the system itself as a whole quote unquote understands Chinese. Even if the one component within the system does not. Searle said, I call shenanigans on that argument, because what if you just memorize all the instructions, so you've internalized it all. You know everything

you're supposed to do. Whenever you are encountering a specific Chinese character, you see the Chinese character, you know what the response is supposed to be. You don't have to consult a book or anything. You still don't understand what it is you're doing. You're just following the instructions that you know you're supposed to follow. That Searl says, does not represent intelligence. Now, there are a lot of other

objections to Searle's thought experiment. There's a lot of heated debate about the Chinese room argument, and it's all very fascinating, And if you think this is interesting, you should definitely look up Chinese room argument because there's a lot that's been written about it, and it's amazing to really put your mind to it and start thinking about the philosophy of intelligence and whether or not it would ever be possible for us to truly determine if a machine possess

that quality. But there some practical things that we should consider too. Now, again, Musk was worried about machines potentially turning on people. I mean, the example he gave was a robot going down the street and killing everybody. Uh. Deciding to do this not entirely possible. For a robot that has any sort of deadly UH abilities, whether it's a soldier robot or something else, there's possibilities of malfunctions or misidentifying someone as a target as opposed to a uh,

you know, an innocent person. And in those cases you would say, well, that's clearly a programming error. But it's not like the machine is deciding to turn against humans. It's more like the machine is making an incorrect conclusion based upon its programming. And again you might say, well, that doesn't have anything to do with artificial intelligence. It doesn't have anything to do at least intrinsically with this concept of AI. UH. And if you were to create regulations,

how do you regulate that? How do you regulate artificial intelligence? Where do you put the limitations? What? At what point do you say, don't let computers do this? Because if you cannot define the problem, how do you create the limitation to prevent the problem from happening? And a lot of people argue that no one is able to really

define what this problem is. People are worried about an abstract concept that they cannot define, and therefore there's no way to create a regulation that is remotely ah relatable to the issue. If you can't define the problem, you cannot create a solution to it. Some people point at different problems, not an existential crisis where robots are seeking us out and turning us into fertilizer, but perhaps a few sure where automation itself is taking away enough jobs

to cause massive economic crises. And there's been a lot written about this over the last few years. It seems like every month another article comes out with either a terribly pessimistic UH prediction of how many jobs will be lost due to automation within the next five years, or a completely optimistic point of view of how many jobs are going to be created as a result of automation

and therefore people are going to have better jobs. Those who are all four automation say the jobs that are going to be UH phased out by automation are going to be the ones that people don't want to do in the first place. They're gonna be the dirty dangerous and dull jobs. So jobs that are either repetitive and

are not interesting and therefore no one wants to do them. UH, jobs that put people at risk, and therefore it would be better to put a machine at risk because you can replace a machine, but you can't really replace a human or the jobs that are just not they take too much in human effort to do uh, and the payoff does not equal the amount of effort needed in

order to complete the job. There are others who say, well, automation is gonna go to jobs that are the easiest to automate, which are not always going to necessarily be ones that fall into those categories. And then you've got people who maybe in an area of the workforce where they don't have the training or education to pursue jobs that are at a higher level necessarily. Their counter arguments

to this as well. Some people say that automation will create more jobs because they'll create more opportunities, with the example of say something like the automation of an Amazon warehouse. One of the arguments is that automation will bring prices down. As prices come down, people will buy more. As people buy more, these warehouses will have to get bigger. As

the warehouses get bigger, more humans will be needed. Even though each human will be responsible for less stuff, they'll be such a large demand for things that that will more than compensate for it. This argument is based off the Industrial Revolution, when the loom was created and people were starting to realize the potential of the loom to

speed up weaving quite a bit. There were real concerns that loom was going to plunge the world into poverty because all these people who had been making a living weaving would suddenly find themselves out of work. The truth is that there was a much greater call for weavers. Because the price of woven materials began to fall, more people began to buy them, and then there was an increased demand for the very thing that people were afraid was going to become a rarity, and so people became weavers.

It's just that they were weaving with looms instead of hand weaving. So there are those optimists who say this revolution with automation is going to be the same thing. Others say no, because it will happen way too fast. Automation is going to change so quickly and so dramatically the world that we will not be able to react to it in that same way, and we will be

plunged into an economic crisis. People like Elon must have argued that this means we should probably look at something like a universal basic income, where everyone is guaranteed a certain amount of money per year by the government so that they can live. They can they can meet their their necessary requirements for the basics of human existence, like food and shelter and clothing. Those who can still have

work will be able to afford more. This leads to some saying that Mark Zuckerberg's response of hey, automation and AI is going to provide lots of amazing stuff for us is accurate, but only for people in Zuckerberg's class, people who are already at a level of wealth and privilege where they will be able to enjoy those benefits because their jobs are not in danger of being automated

the way other jobs are. So there is a different crisis that's on the horizon, according to many people, and it all has to do with automation, not just artificial intelligence. Automation doesn't have to incorporate a whole lot of AI in order for it to be a threat to jobs.

But the verdict is still out as to how great a disruption it really will be and whether or not people will be left behind, or that we truly will find new jobs for people, new opportunities will arise that will end up being superior to what they would have done otherwise. It's an unanswered question, but it's one that

a lot more people are asking. It is not directly related to what Musk and Zuckerberg were bickering about, however, So I was looking over some information about various jobs that are have the potential of becoming automated, and I found one in the Atlas dot com that was pretty interesting. According to this research, the jobs that have the most potential for being automated include accommodation and food services at

a whopping se potential for automation. Now that does not mean that all those jobs will be automated, but it does mean that it is the most likely out of all the different categories listed to undergo automation. Other ones that are high up on the list, including manufacturing, transportation, and warehousing, which we're seeing with Amazon. Agriculture retail trade

is at fifty minings. At the ones that are at the lowest end include educational services at management at thirty and boy howdy is that going to cause a rift if that is true? And then there's a vague category called professionals at I'm assuming professionals means people who are working white, colored jobs that have a lot of variety to them, and not folks name leon who go around with pistols. That's a leon the professional reference. So is Musk right? Are we going to see AI rise up

against humans? I don't think that's going to happen in the near future. I think AI does pose some challenges, and if it is incorporated in ways that we don't fully think about, it can cause at least short term harm, if not long term harm. But I don't think it's an existential crisis. I don't think it's something that we need to worry about regulation at the moment. Zuckerberg's argument that's going to improve all our lives, I don't quite

buy that either. I think that it will uh have minor impact on most people's lives from a from a direct perspective. If automation does end up affecting more people than obviously that's a negative impact. I think they're both slightly wrong, and there have been some writers who have suggested that perhaps this argument is not really about AI, but more about Zuckerberg and Musk both promoting images of themselves. The argument is not really about artificial intelligence. It's about

I stand for this. That's what my reputation is based on. Therefore I need to continue in this vein. And I find it particularly interesting that Musk is talking about AI being this potentially dangerous situation, since it is and a very important component of both Tesla and SpaceX, and so he's walking a very fine line. It's also an important component of his proposed tunnel system from the Boring Company, which is all about earth boring machines, not things that

are dull. So it's an interesting debate. I'm not going to get involved in it online because neither Musk nor Zuckerberg know who I am, and honestly, I think I prefer it that way. But I'm curious to hear what you guys think about this AI debate. Do you think that we're in an existential crisis or heading toward one, or do you think it's much ado about nothing. I'm curious. Let me know, send me messages. My email is tech stuff at how Stuff works dot com, or you can

always drop me a line on Facebook or Twitter. The handle of both of those for the show is tech Stuff hs W. Remember you can go to www dot twitch dot tv slash tech stuff to watch me record these episodes live. You can chat with me in the chat room, and just go to twitch dot tv slash tech Stuff. You'll find the schedule right there. I record Wednesdays and Friday as I hope to see you in

there and I'll talk to you again really soon. For more on this and bothands of other topics, is it how staff works dot com