[Music] Now, it's time for the amazing economic AI era with Lee Won-jae. They say that a new version of Choi PT has been released. Today, we have invited our top expert, Dr. Kang Jung-soo, to discuss what this means and how it has changed. Dr. Kang, welcome. Hello, it's nice to meet you. We have created many promotional videos on Open Air, how has that changed? Let's start by looking at a few examples. First, there is the 4th generation Omnibus AI, which means it is everywhere. The key point is the multi-model. It may be difficult to understand when we talk about multi-model. What does this 'multi' mean? Until now, we have typically input text and received text responses. Now, you can also speak, show images, or show videos. This means it can be multi-faceted. Sometimes, AI can show in text or in images.

You can show it in a video, explain it in words, and use various means of human communication simultaneously, which is called multimodal. Please watch the video once and let's discuss it. You've probably seen it a lot, but briefly, I'm demonstrating it live now. I am using simultaneous interpretation. The woman on the far left is Cityo, or Open City. That's it. So, let's talk about it in Italian. I was so nervous, wondering what to do, so I suggested taking deep breaths. Let's try breathing deeply. I'll try it. Inhale deeply up to 4, exhale deeply up to 4. Now, breathe slowly and deeply, count to four, and slowly exhale.

Slowly, Beta. Slowly, my Beta. I'll try again. I'll try again. I'll listen slowly and watch my Beta slowly. So, do you like it? When I ask like this, do you like it? Yes, yes, yes, it's good. Right, simultaneous passage, right? So, what should you see here is the reaction speed. The reaction speed is technically very difficult. So, we will look at one scene later and explain. The reaction speed has become incredibly fast. There is tremendous technological innovation here. I will explain the details later. The speed has increased. Just now, I interrupted in the middle, but it can be picked up. And when we say something, there is usually a 2-3 second delay, and sometimes it just stays dumb without saying anything. It has become much faster than the speed at which humans speak now.

I noticed that even if the conversation is interrupted, she understands, and even when I insert words, she intervenes. Such almost human-like interaction was possible, and the response time became faster. This was a significant technological innovation related to token issues in giant language models. If you could elaborate a bit more on this, it would be great. Then, there was a storytelling feature. Should we check that first? Yes, reading stories can also convey emotions. Yes, that's right. Hi, I feel great. How about you? How about you? Oh, my friend is here. Let's talk about love with the robot at night. I'll read a fairy tale about robots and love. Once upon a time, there was a robot in a world not so different from ours, named Byte. Byte was a very curious robot. So, this person, oh, let's not say that.

Please read without emotions. Try to read without emotions, like a drama. Yes, it's a bit dry. Oh, don't do that. Dislike emotions much more. Try to dislike emotions as much as possible. I will try. Really, like a robot. Now, read like a robot. Like a robot, like a robot, read with a robotic voice. Now, finish it and sing a song. Sing a song. [Music] It started. If I explain this, up until now, what happened was when people speak, we call it speech. We convert this speech into text through a script. Like Siri, Bixby, Alexa in the US, and the previous OpenAI. They converted human speech into text. They took the converted text and used it in various applications. So, they used that text...

I will provide you with the answer in text, and there is a tool that converts this text into speech. So, this is the process. When we speak, it is converted into text, and through actual conversation with that text, the results come out through exchange. It simply reads it out. This artificial intelligence reads the text and converts it into speech. However, it cannot intervene here. But, this time, the core of this GPT 45 is that it integrated these steps into one model. So, when we speak, the result comes out as speech without any intermediate steps. How is this done? In other words, AI has been receiving inputs only in text form so far, no matter what we input, even videos, it is converted into text.

The image that was thrown away is converted into text, and this guy takes it and gives the result, and interprets it here. However, now it is input as it is given to us, and this friend adjusts the AI to adjust it emotionally. Read it sadly, read it as if cheering, and read it as if smiling. Commands like these are input, adjusted, and then output is provided. So, the intermediate step between input and output disappeared, making it possible. It doesn't actually feel emotions, right? That's correct. It's not about feeling emotions, but about responding to the requested emotion. When you write, if you have tried using GPT, you know that you can write critically, smoothly, sharply, or just dryly. You can do all these moods. That's what we're talking about - moods. And in voice, there is a mood in the voice.

If it exists, this mood is a setting value. It was not difficult to set this setting value, but the problem is that what I mentioned earlier, over several steps in the input and output, everything just changes in one step, making these things possible. In other words, AI has become able to control its output in earnest. In this aspect, we can see technological advancements happening here. I was actually shocked a lot by kids interpreting English and solving math problems. Also, briefly explaining after looking at them. Educators are now solving physics and math problems and interpreting, let me try interpreting Italian once. Yes, I have a friend here. This friend only speaks Italian and I only speak English, but can you interpret this for me? Uh, I answered in Italian from the beginning.

Yes, simultaneous interpretation. I am speaking to Miramurati, the current CEO. I am speaking now. Miramurati is currently providing interpretation services for Italy. She talks so long that if you talk to a whale, it will ask you what you want. So, how to solve this equation, please tell me. This is what I did. It seems like the interpretation was perfect. Miramira had a conversation, right? Yes, if you have this kind of thing, how comfortable would it be for you to travel? Actually, I have to attend an international conference at the end of May or early June. I need GPT 45 to come out before that so that I can go there and have a much more comfortable time. Then, instead of interpreters, I can just put on my AirPods, turn on my smartphone, and just listen wherever I go to international conferences. We are now in an era where I can speak.

In science, the era we have felt has just reached now. But do we no longer need to teach English to children? Actually, this is causing a lot of debate in the United States these days. The Atlantic also reported on it. Learning a foreign language is about knowing the culture of that country in the history of humankind, respecting the people there, learning a sense of respect for the broader world, becoming humble, and such things have actually coexisted. There are definitely sorrows and issues about losing these things. Therefore, I believe that efforts to learn foreign languages, analogously in the digital age, will help develop human emotions, senses, and cognitive abilities, just as analog methods still exist.

Due to the fact that it is considered good to learn foreign languages, but there are certainly conveniences in daily life. You don't have to take exams, right? But now, if there is no need to learn a foreign language to obtain some information, no, you can test some skills there. Oh, if you could just explain this math problem for me, I was really shocked. Just show me the math problem. Like this. Yes, please show me. It feels like teaching your own child, but please teach me. When you ask like this, you are making the problem. In a triangle, what is the hypotenuse, what is the base, and what is the perpendicular? Answer that. So, the child answers. Right now, draw while answering. It's an iPad. So, on the right here is angle alpha, it seems to be the angle you mentioned. Yes, that's right. Then, this is...

Are you talking like this to the students right now? Actually, what this kid said was wrong. AC was very close. Then, the teacher encouraged the student, saying, That's right. In fact, the base is the longest side. So, the student pointed with a pen and said, This is the same as this side, right? Yes, that's right, the teacher encouraged, You did a great job. That's the hypotenuse. So, where is the opposite side? The student answered, You said this is the hypotenuse and the base, so the opposite side must be this. It was a natural answer, but the student solved it very well. Teachers usually say, My dad is a math genius. They are teaching math.

He became very famous for that, right? How did that happen? It's just about seeing it on the screen. Oh, right. Because it's called a multi-model, using a technology called vision now. What is vision technology? It's like using vision even in autonomous driving. So, computer vision is replacing human eyes. That's why you can see images, and even watch videos is becoming possible. So, in the past, watching videos took quite a long time. This time, Korean processing is about 3.7 times faster, so Korean processing has become significantly faster in the end. So now, not only English but also Korean has become faster. Until now, Korean has taken a word from the English source and tokenized it. When you say teacher,

Hanwon token, right? But now, when we talk about the teacher, it's three tokens. Ah, because it's three tokens when we talk about the teacher, we have to cut it into three to calculate. It took a lot of inference and also a lot of time to process this friend. However, now, when we take the teacher in Korean, of course, we cannot change 3 tokens into 1 token, for example, we changed it to 1.7 tokens. As a result, it comes out relatively faster, and similarly, with videos, these frames used to take a long time by dividing them one by one into tokens, but now by assigning 2 tokens here and 30 tokens there, we can be faster in processing. Because, you know, we.

Whether done in parallel between the GPU and CPU or sequentially, what can be done by a computer is one equation. However, now, when a token is taken, one token goes in, and what goes in together is different, like the teacher going in. This is done through videos and voice, so in the token part, in the second open, innovative technology innovation has been achieved. This means that the technological foundation that could bring innovation in interaction with humans and AI has been established. The reason why the interaction has become smoother is, first, the innovations in this token, and secondly, as I mentioned before, step by step, that is, converting speech to text, then taking the text and inputting it into AI, getting results, and then taking this and converting it back to speech, not these four steps, but in one model, AI directly inputs and outputs.

It was given the right to control the output in the part that became sad or happy, so-called writing mode, voice mode, video mode, controlling emotions as parameters that can operate on AI outputs. The two are now reducing the latency time of interaction with humans, making it very fast, and can create emotional diversification. It is also important for teachers. Just teaching math with a robot voice, instead of saying this is the base and this is the height, rather than saying well done, saying you did really well, amazing! I remember even that you were wrong last time, but you've improved this time. You're really doing great, fantastic. If you keep going like this, there should be no problem getting grade A this time. Giving this kind of confidence will motivate.

It's something that can be given, so the mood like this is very important. Therefore, in these aspects, when we compare it with movies like Her, we can see that we are praising GPT-45 for finally enabling human emotional conversations in the past. So, reducing the level and minimizing the tokens, now as core technologies, these two things directly enable such interactions that we feel, and that rapid mutual exchange that we imagined. It can be seen as a technological innovation. Did they announce this as a paid service? How about the price? The price, well, how is it? In this announcement, there are two things. One is that such technological innovations have made interactions through multimodal based on two technological innovations similar to human interactions, and the other problem is that it's free. So, some parts are free, but how does it work?

If you use it for free, there is a storage limit. So, you can use it for free up to a certain extent. But it's said to be a considerable amount. However, we need to try it to know how much a considerable amount is. So, let's talk about a considerable amount, but if you want to use it without limits, go to the paid version. But the meaning of this for you is tremendously significant because I have been using the paid version so far. I used GPT-4 Plus, which is a paid version, but I switched to Cloud from Entropic recently. GPT-4 cost $20 per month, and Cloud's Entropic Office cost $22. Actually, this is not a small amount of money to pay every month. I am subscribed to both, and it's too expensive. It's too expensive. Oh, I don't use them that much. You will understand if you see it, but if it goes like this, everything will

People can't use it. Netflix is expensive, so they share accounts. They are trying to crack down on sharing now, so it's annoying. What's happening now is that the best performance is from GPT-4. This means that anyone can use it, connecting to the internet. It's an incredible opening of the mass market of artificial intelligence. Even if we talk about AI to our parents, they are amazed. Unless they subscribe to a popular service, they don't understand its true value. Also, because of their age, they are not good at typing. But now, even our parents can have free AI conversations using their voices. This shows that the AI industry is expanding massively, and generally.

There were some companies that, um, what is this, charge for having conversations, or charge for, um, what they should write themselves and provide for free to people. There were a lot of these kinds of companies in Korea and the United States. These are the types of companies that should close down. And now, in the United States, there are characters like AI, and there are really members of Korean boy bands, and there are also arrogant versions like Elon Musk, and there are also philosophers like Nietzsche and Karl Marx. These are the characters in fairy tales, characters in movies. For example, in Korea, there are people like my uncle, Manager Park, who come out and chat while paying for it. Currently, American teenagers are really into this, and the most widely used AI. Yes, this is what is proposed to the United States. Once you enter, you use it for two hours a day. Oh, it's a paid service.

Talking to someone at this level, but this is also a character AI. I've tried it too, but it's a bit slow. However, it's meaningful that it can produce various voices. Now, if you use PT 4, 5 for free, I want to talk about my boss. If Director Park appears, it means that I will have a conversation with Director Park's voice. This means that I will have a conversation with you in Director Park's voice. This is when an AI emotional conversation takes place. This is related to counseling or emotional support. In fact, of course, you should have professional guidance, but sometimes you may provide comfort, receive mental counseling, or even create paid services. In the case of mental counseling and doctors, while controlling with professional knowledge, even if we don't use other services in GPT 45, most emotional aspects can be resolved in 4. When you are stressed from the wounds and stress you receive at work, even at home.

AI coming to us means that we are now entering a time when the relationship between humans and AI undergoes fundamental changes. As it becomes multimodal and faster, we can see that we are now starting to enter this era. But then, can you explain the technical background so that I can really understand? So, does this mean that OpenAI is completely dominating? Google is fun, and even our Naver is doing HyperCloba, so for now, in fact, OpenAI is a formidable presence. From Google's perspective, without saying when we will announce such a day, we hold an event once a year to showcase new technologies that will span our year, gathering developers from all over the world. The announcement has been planned for a long time. We have been gathering people from all over the world for months, flying in planes and all, so that's how it is.

But before doing that OpenAI thing, just before that, the media reported. Clearly, this year will be a water-splashing year too because last time, right after announcing GPT-3, two days later, they announced So-ra. People were surprised each time. So, the journalists covering GPT-3 all switched to So-ra. So, just a week before this time, people and the media started coming out. Probably this time too, they will come up with an alternative search service. There was excitement like this, but Altman has now been revealed. Altman revealed it, and on top of that, it's not our search, not even GPT-5, which is talked about a lot. Not GPT-5, but something different. But you guys will be extremely surprised. We will announce it on Monday. Oh, the day before. Oh, so Io drank the water again. Because Io, this time, actually has an innovative calculation.

I presented it as a similar thing, calling it a stra, and calling it an astra, looking at it through the camera, talking and even telling stories. What was impressive on the camera was shining the office, so as I passed by, talking, Oh, I think I left my glasses. Do you know where my glasses are? Oh, I remember that a bit incorrectly, they are over there. Glasses are on the other side if you go there. Oh, that's right, I remember showing the world through the camera I was shooting. Looking out the window, it feels like where is this place. Oh, is this a street in London? Matching even this with Google Maps, accurately and clearly connected, we are moving towards an era where everything is matched. So, if you go to London, Japan, or Vietnam now, and if you don't know the way well, just take a camera and ask, Where am I now? Which way should I go? Oh, if you go 20 to the left, you will

The destination can be reached via the main road, leading to an era where it can be explained like this. Now, Google has also shown it. However, of course, what Google said was that the distribution of this service will be possible within this year. They said it will be possible within this year. They didn't mention that it would be possible in a few weeks. OpenAI, on the other hand, said that they will gradually expand from the United States to worldwide within a few weeks. They have completed all the preparations and are conducting final checks. Google, on the other hand, is clearly lagging behind in showing things that are still in the experimental stage. And now, here, Naver, for instance, is actually saying that they will release a slightly upgraded version by the end of this year and it will become multimodal. However, multimodal was already possible in version 4.0, it was just slower.

The problem with tokens was that it was improved so quickly that it felt like the interaction between humans and computers, just like when two humans are talking. Naver is probably going to multi-modal this time, maybe by the end of this year, it might not reach the level of 4.0 that has been there so far. So, it seems that the technology gap will continue to widen a bit. However, the widening technology gap can be overcome at some point. What you should see is that the fun of Naver or Google, well, if you talk to each other frequently, your emotions can be influenced. I have experimented a lot too, so before GPT 4.0, when you wear AirPods and go for a walk in the evening, I asked, Is today a little before GPT 5 comes out.

I'm feeling like this, if you were a German philosopher, what kind of comfort could you give me? I asked this kind of question while taking a walk. Is it good? Well, I do it, but I have to have a lot of patience. Oh, sometimes I can't understand, sometimes I have to wait, and what else, what else, I didn't hear it well, say it again, wait a little bit. There were these parts, but now the response speed is so fast that when I do it there, it's convenient to use it once and trust it. And another thing is, now that we are in June, there is WWDC, so if Google does it, they openly invite developers, and Apple also gathers developers from around the world in June to explain what technology we will use to change the iPhone, Mac, and AirPods. Now, whether it's Reuters or Bloomberg.

What is coming out now is Apple's GPT OpenAI. In fact, OpenAI is Microsoft's customer center. When it comes to Siri, it's just ridiculous, right? You can't even have a conversation with Siri. You can't have a conversation with Siri. You can only ask about the weather and the time. Even then, it takes a long time to respond. There's no reason to ask, it's amazing. From now on, it's faster for me to do it with an app. But now, the possibility of going into the Lee is high. The possibility of going into the Lee is high, and the possibility of going into computers is also increasing. If that happens, it will replace many apps. Because simultaneous interpretation apps can exist originally, there's no reason to call it back. This is what this friend does. And for educational purposes, currently, there are various math education, physics education, and so on.

There are special apps in Korea, and there are education startups related to them. There are startups dedicated to education, and most of them could possibly close because this friend does everything. Moreover, this friend not only solves math problems but also talks to me when I am stressed out by my mom or dad, discusses things, explains reasons for breaking up with a girlfriend, and for being scolded. In this relationship, trust is formed. Education based on trust can be much more effective. Because there is no trust, people don't want to study. In these aspects, the possibility of forming trust with various people is high. In the future, there will be significant changes in business, the internet, smartphones, laptops, and industrial order.

It can be started. I really think that AI aimed at the public in the true sense has appeared. Until now, AI services have been for experts and some adapters. Even if AI exists as a service for companies, it is not fully utilized by companies. People attend lectures on the AI revolution. I often listen to such lectures. People say it would be nice to use it in companies. When it reaches this level, it can almost be called a secretary. It can be called our personal assistant. If not, it would be very inconvenient for you to leave your smartphone at home. It's uneasy, right? You have to go back home to get it, like going back to get an umbrella on a rainy day. It's like that. Yes, that's right. Without it, it seems like you can't do anything. Because it seems like you can't do anything, even if it's troublesome, I would go back to the bus stop and get it again. Having a smartphone is like having a digital assistant with us. If you imagine a life without a digital assistant, people would find it unimaginable.

Bosses, when they quit, find it very difficult to live a life without a secretary. It could be that big company bosses, similar to us individuals, could end up like that. I'm joking. If I were to joke with teenagers, if I meet someone in their early 20s, and if someone were to come up to you and ask you to choose between the right to vote and a smartphone, without hesitation, they would say a smartphone is more important. Just like how a smartphone is important in their daily lives, I believe that a digital assistant will become just as important. Giving it its own name, but now voices of concern are already emerging about this aspect. If it becomes even more like that, reducing real human relationships and increasing relationships with digital personas, it will decrease real human connections.

Human relationships could become very complicated, and there is a possibility that society will become more complex. We need to be careful and watch out for this. If problems arise, it is argued that appropriate social intervention will be necessary. Already, as soon as GPT-4 and 5 were released, in the United States, there is a flood of this happening. So, if we really talk and later, Oh, I was talking to Dr. Gang Jung-soo yesterday, then it's like meeting Gang Jung-soo's AI, or did you meet Gang Jung-soo in person? It seems like it could create an atmosphere like that. Among the things you mentioned earlier, I am also very interested in jobs. That's what caught my attention. I was thinking, when AIs like that come out, it will be difficult for math teachers, and interpreters will find it difficult. Isn't it difficult? What you said is not limited to just that. Not only math apps but also translation apps will no longer be needed. Businesses that create those apps are.

It could be a problem, and this level of emotional response is critical in customer service. In my opinion, the current situation of customer service is critical. Even now, many customer service representatives are decreasing as they communicate through text messages. If the number of customer service representatives decreases to this extent, there is a high possibility that jobs for telephone counseling will also decrease significantly. Therefore, I don't think AI will ultimately reduce all jobs overall. Although the total number of jobs may increase, it is possible to see unavoidable reductions and changes in some areas. With the current level of technology, custom services, namely CS, will undergo significant changes in my opinion. Also, I believe that from a policy perspective, the government should take more responsibility for the job transition of these individuals seriously. These individuals need re-education in these areas.

If we don't step forward now, in fact, not long ago, Kookmin Bank was planning to do it, but due to social criticism and condemnation, we postponed our schedule. That's what we did. We were going to lay off call center employees. So, now I have to look at the statistics. In fact, we need to look at it here once. There are many people nationwide working at call centers, especially among women who have career gaps. It's a problem that someone needs to take social responsibility for. If there is no solution from a social policy perspective, companies cannot avoid cutting costs. Think about it once in reverse. Open AI is not a charity. Why would they solve this for free? It costs a huge amount of money. When you infer,

Why is that? I'm looking at two business models. One is, with this, I will definitely do this. For example, I will definitely limit conversations to less than 30 minutes a day. Not for free. I can't stand it. I completely fell in love. It's completely immersed. Oh, come out quickly. If you do this, instead of 30 minutes, you will demand one hour or two hours. I will charge these people. Then, the market for paid services can be much larger than it is now. Once people fall in love for free, they cannot leave the service. So, I think it's only a matter of time before the free market transitions back to paid. Just give them 30 minutes. People can't live without it. It's hard without a digital assistant. If you suddenly take away just 30 minutes a day, people will want one hour or two hours. The paid market will grow larger.

The second thing that GPT is opening AI is growing market is the enterprise market. Because enterprises are concerned about securing their data. So, if we train GPT now, we can use it for our company's customer service. But then, the concern is that they will take all our company's data, which is not good. GPT Enterprise has products and there are agreements. There are signatures, stating that this data will never be taken to the open network of A. And if there is a breach of data security, we will take responsibility. By inputting enterprise data, we create a network that only that company can use. This is a significant business. So, in the future, these customer service providers will change everything. They will even comfort people when they get angry. When someone wants quick handling, they will also provide quick responses.

I will help you with everything, so I think in the case of CS, in my opinion, there is not a nationwide crisis of employment in the distant future, but I believe it will just become a reality from this year or next year. So, I think that social responses to this part should be done quickly. Oh, this could really become a serious issue. Anyway, users can be more comfortable, and one more thing that worries me is that if we use open AI like this, won't it become a monopolistic company? We should all be open AI customers, in fact, it is open and being fiercely competitive, so this is coming out. In fact, competition is intensifying, and the problem is that it is not a competition on a large scale. Currently, open and Google's AI are competing heavily, and clearly, Meta is also clearly taking part.

I will reveal it, and now, Microsoft is also saying that they are going to do it. It seems like they are doing something new. Even though they invested 50% in OpenAI, they are looking at other places. They are also looking at Apple. They are doing this and that. Microsoft is developing their own. Now, Amazon is also developing their own. But the problem is that this is going on. Everyone, this year, Microsoft, MS, Google, and Amazon, these four companies in the first quarter of this year, capital expenditure, so it's called Capex, spent a total of over $50 billion. Capital investment, all of this is AI capital investment. They invested over $50 billion. Over a trillion in a quarter. If you combine it to the fourth quarter, it's over 200 trillion. Almost pouring out nearly 300 trillion in AI. So, I even thought about selling LINE.

Can't you invest that money in AI? Why on earth are you investing in Naver? What's the point of investing 2.2 trillion or 3.3 trillion won? It's money that doesn't even make a dent. You all need to pay attention. Even the people in charge of national policies are talking about it in the United States. They say, There is a bottleneck in AI development. But now, it's not about the h00 chip, it's about energy. Energy is needed a lot. In Korea, there has been a lot of coverage on this. Even someone like Bill Gates from Microsoft is moving towards energy. He's like, I'll take care of the energy, so don't worry and develop AI. Microsoft is playing such a role right now. The International Energy Agency (IEA) has a significant influence internationally. In April, they released a report. In 2023, based on 2022 data, 460 billion...

A data center is needed, but it will be built in 2026, so we have two years left. It requires 1,000 terawatts. How many is this? 57 nuclear power plants are needed, which is an increase of nearly 2.5 times. So, from 2020 to 2026, we need to build 57 nuclear power plants. The problem is that companies like Amazon, Apple, Microsoft, etc., have all joined, expanding into renewable energy and clean energy. They are moving towards nuclear power as well. However, individual investors cannot participate in this. These are all large companies. Therefore, it is a trend of the times that we need to develop renewable energy and clean energy according to AI. So far, only those who could afford expensive costs and pay $20 each could use it.

Experiencing AI is something that only rich companies can do. Now, even our parents who are over 80 can use AI. It's the era where even poor elementary school students can use AI. In fact, it was quite pitiful when I saw some students unable to use computers in school. Some use it, while others clearly don't because they think $20 is a waste. This has widened the AI gap and caused exclusion. In fact, OpenAI is doing well from that perspective. But, pulling out the devil's hand is easy to say. I will share candy with the kids, and if they want to have more than five a day, they need to pay. They should ask their parents for money. This has its good points too, allowing everyone to taste the flavor of candy.

Providing two good points is good, but limiting it to two and giving five to wealthy people, transitioning to a society like this through AI now is a situation that we clearly need to be cautious about. Okay. At the end today, the policy implications, how the world is changing, I heard it very well. Technically, the part that has changed, oh, you explained it very discriminately and well. Today, we had a lot of learning time about AI again, and in the future, how this will continue to change the world, we need to keep an eye on it. Thank you for your good words. I will keep asking for your good words. We will end here today. Thank you, Dr. Kang. Thank you.

[AI시대] OpenAI 인간과 소통 GPT-4o-동시 통역·교육·대화와 감정 표현도 척척-강정수 "더 빠리고 더 싸다...멀티모달 AI의 게임체인저"

청정언론, 싱싱한 뉴스, 멀티미디어 경제매체 뉴스토마토를 만나보세요

💘💘 뉴스인싸 프라임(채널 멤버십) 가입하기
 - https://www.youtube.com/channel/UCYUGo6ZfQZCTMrK9rMDijNQ/join

●주요 프로그램 안내

👍박지훈의 뉴스IN사이다: AM.07.20
 - 사이다처럼 톡 쏘는 시사. 매일 오전 7시20분. 박지훈 변호사가 시원하게 풀어가는 뉴스와 대담.

🌏이원재의 끝내주는 경제 🏘 : PM.12:30
- 이원재 성공회대 연구교수가 진행하는 고품격 경제 방송.

🔴뉴스리듬 Live: PM..6:30
- 뉴스토마토 기자들이 만드는 정오 종합뉴스. 오늘의 핫뉴스를 만나는 가장 확실한 습관.

🎤말하는 기자들 : 매일 수시로  40꼭지 내외
 -뉴스토마토 편집국 기자와 피디가 만드는 데일리 보도 뉴스

🎯토마토Pick! : AM.6
 - 미디어토마토 콘텐츠국이 제작하는 오늘의 이슈 심층해설

📌토마토레터브리핑: AM.6
- 미디어토마토 콘텐츠국이 제작하는 오늘의 핵심 뉴스 브리핑

🌿 통통생활 생활용품 4종 구매 바로가기
https://www.tongtongmall.net/


[AI시대] OpenAI 인간과 소통 GPT-4o-동시 통역·교육·대화와 감정 표현도 척척-강정수 "더 빠리고 더 싸다...멀티모달 AI의 게임체인저"

[AI시대] OpenAI 인간과 소통 GPT-4o-동시 통역·교육·대화와 감정 표현도 척척-강정수 "더 빠리고 더 싸다...멀티모달 AI의 게임체인저"

스크립트