Before we begin, can I have your permission to record this interview? Can you state your name and your position at Cal State Fullerton, please?
I'm an I'm an associate professor of computer science department.
Okay. Can you spell your last name please? So, it's
filling garden P A, N A N G, A B, A N.
All right. Thank you. As we discussed before, I'm going to ask you a few questions relating to the AI so my, my first question is that how can I understand AI like, what exactly artificial intelligence is in terms of technology like what it is what is that? Just one minute.
Noticed
so, so formally, Korea is basically any technology that can operate somewhat independently and at a level that is comparable to a human, you know, or some other kind of living organisms.
So Oh, okay. Are you and continue talking about that? You know, yeah, I
mean, I think it's essentially I mean, it's the ability to, to, I mean, to generate whatever outputs you know, someone in your body call with within, in an independent way. It is it's not strictly programmed to, to respond to inputs in a predefined way. It basically has the ability to, to generate outputs in an in a somewhat independent way, like how humans work.
So how can I help people in life like because it's so convenient, right now today? Right.
So, the main advantage of AI or at least the pope of AI as he reviews in what we may call, drudgery, so you know, work that we that, you know, that we need to do, but we don't really want to do. So, that's the kind of work that we hope that AI will, will do. For example, if you want to go somewhere in a car, you know, we have to drive it. But you know, many people don't want to spend the time driving so that's an example of AI individually See, that kind of work for us. Another example is if you need to travel if you need to book a ticket, you know, currently we you know, we look for the tickets, you know, we have to find out which is the cheapest ticket, then we need to decide, you know, what time do we want to leave and then you buy the tickets and all of that. But if you had an AI sort of helper, if you can imagine that that AI helper will, you know, will do the entire booking for hours so that we can we have the time to enjoy the travel and not worry about the travel arrange.
So, scene is so like really helpful. I agree with you that it can help humans live like make our lives like get easier. But I think like one, there's always two sides. of a thing. So what is the bad side of AI?
Well, there's two, two sides to it. You know, one is negative sides of AI itself. And the second is the negative side. So what happens when AI is used a lot in society. So right now, one of the negative sides of the AI technology itself, like right now is that the AI technologies themselves do not know if the answers that they're providing or the actions that they can take are for example, correct or fair. There is no way of checking that right now. So for example, if you ask an AI a question, the an AI like you know, for example, chat GPUs and answer, chat up it is not guaranteeing you that that answer is correct. It's just giving you some answer and it is up to us the human to now to figure out the answer is correct or not. And somewhat related to this is the idea that we really have an aim for a question in the recommendation are asked to do an action. Current AI systems do not have ability to know if the answer is for example, fair or biased. So that's another downside is that it's not capable of, you know, showing these human values of being fair or unbiased.
Yeah. When you say about the chat GBT, I feel like right now, I there's an article that I read online. It said like, right now there's over like, 200 books being sell on Amazon. And mostly it's written by chat GBT. So that that is something I'm really concerned about. Because right now like, it's gotten more intelligent. Right now. It's not stopped. Only writing a few sentence is writer can write a book. So do you think like it can replace the creative writing in the future?
Um, I don't think so. So I think it will replace some kinds of writing, you know, like, you know, maybe writing reports, you know, maybe automated, but I think one of the, you know, one of the things that we like about creative writing, is actually the writer itself. Right? So, for example, when people say that, you know, I like this particular writers book, they're actually sort of saying that, I like the writer, you know, it's like, the reason why we value the creative writing, is because it comes from a human, you know, and we know that this writing is an expression of the human six, you know, the humans experience of that human persons hope, you know, you know, so already that humans secrets. So if, if now this creepy writing is done completely in an automated way, you know, none of that is true. You know, so for example, if you read a book and we read a book about, about love, you know, that is interesting to us, because we know that the writer in a human has experienced love and is writing about it. Now if if an AI system like John GPT is writing about love, it's not talking about any one person's experience about love. So, you know, it's less interesting, I think. However, if we wanted to, if we want to write a report you know about, about love. You know, I think it's okay for an AI to write about it, because the AI is in essence summarizing everything that people have written about love, you know, on the internet. So I think they're looking at, you know, like the, what we want from Creative Writing is little bit different. From what we want from say, like reports. And I think one of the important things about creative writing for most people is that is that there is a real person behind.
Yeah. Well, I think like when you say, the checks UBT can like somehow affect the report reporting? It is somehow I concern the most because my major is journalism. And some people said it can replace like journalism in the future. In term like, I one time I asked church GBT and can you write me an article reporting on the earthquake in Turkey? And it immediately like, I think only five or 10 seconds and it gave me like, a short article on that. But unfortunately, what what I read from it is that they it made up a lot of sauces. It made up like fake people who like answer the interview, and they don't really give a Pacific like the update on the earthquake. So it kind of limit on the knowledge that GBT can get right now. It can be like, provocation, like a lot of that. But it doesn't mean like people don't rely on chat GBT to get like information to writing something. So how can like AI or GBT in general can affect human if we rely on it too much.
Yeah, I mean, I think that's an important point. I think this is related to what I was saying. Is that right? Now AI, especially things like champ GPT have no way of checking their hands. So I mean, does this mean GPT? I mean, if you ask that PvP I mean, is this true if there is no way for GPT to check if it is true? There's also one other small relationship with Chad GPT is, you know, these are all what's called machine learning systems. So they were trained on data that they found on the internet. And Shabd GPT, for example, is trained primarily on data that comes from before 2021. So if you ask about the incidents that happened recently, like the turkey earthquake, then it's not you know, it doesn't even have that much information to answer. So it's just going to make up make up things. So yeah, one is there is I mean, one issue is that AI systems don't know that it's true, but I think your question was more about if everybody uses AI, in what negative things can happen. And so I feel like the idea of the AI not knowing the truth, you know, hopefully that will improve but I think there is a bigger issue, which is that, you know, you know, humans, you know, we have, you know, we you know, we have the ability to you know, to make exceptions, but you know, we sometimes intentionally break rules because we feel that, you know, the rule is not it's not correct to apply the rule now. I'm right, I mean, AI is harder for it to know when to break rules. So you know, AI may know the right rules. And a good example is in a driving a car and then the AI can learn to do and, and learn the rules about driving a car, you know, where to go, how fast to go, when to turn and so on. But if an if something unexpected happens in Russia, it's not clear in how a how AI should respond in a human's view, respond, you know, based on a set of values that we have, for example, you know, the way you drive a car will depend, you know, whether your child is in the car with you or not. Yeah, so, things like that. It's like you know, maybe you get into an accident with with a person, you know, the way you behave, you know, in some sense, it depends on who you got the accident, you know, like, for example, did you strike up a pedestrian or bicyclist? Did you strike an old person really strike a child, you know, the way you behave will be filled Bill, you know, will vary, but it's not clear if AI can have the same set of reactions to unexpected events. And if those reactions to unexpected events will be based on human values. So, one example is, I mean, it's just in the university. I mean, I've said that in most university courses, you know, the, you know, the instructors will say that they will not accept late assignments, but reality is, you know, many instructors will accept late assignments because student approaches them and you know, explains their situation, you know, and depending on the situation and depending on the instructor, and depending on you know, several other factors the instructor may or may not accept the late assignment. And that is because, you know, inspectors are also human, but if you had like, let's say AI instructors, you know, the AI may have, you know, a strict policy saying that no late assignments and you know, will absolutely not make any exceptions, and that may be a negative thing in the long run.
Yeah. So, I have one final question for you, since you are the professional on the campus. about technology. So my question is like, what you're seeing right now, so do you support or against the AI?
No, I think I mean, I mean, I have high hopes for AI. Because I think one of the things to remember about AI is that we need to compare AI with the cost of not using AI. So again, an easy example is self driving cars. So it's possible that self driving cars will get into accidents. You know, that may be true, but the fact is, you know, humans driving also create accidents. Yes. So, so, we should not say that oh, it was a self driving car can create accidents. So we should never have self driving cars. We should also compare it to the cost of, you know, not having self driving cars and you know, several people die in car accidents every day. So, I'm actually hopeful that you know, the AI will be beneficial. And you know, but it will not be a smooth ride and as we use AI, we will see some downsides. And then we will need to, you know, correct the AI technology. We will also have, I think you know, more regulation, so that, you know, if, like if something goes wrong, you know, maybe you know, you know how to you know how to compensate the people because right now, when a human gets into an accident, we have this insurance system, you know, that pays for the you know, that pays for medical bills and things like that. And I think we will similarly have systems that will work around the weaknesses of AI to enhance I feel positive about the AI prospects. Yes.
Okay. Thank you so much for talking to me today. And Sophie, she had that. Yes, well,