Hello, in the Magnify Series, we ask the experts the questions that we want to announce the answers to in the field of growth, our guest today is Erdem Eser Ekinci, the co-founder of galaxy and dp, who has the vision of establishing an artificial intelligence company at 2009. We have hosted many of our friends here so far. We asked a lot of questions, and in all of them, the topic came up with artificial intelligence. Endured. I am happy that we will finally be able to ask a question that I want to ask a lot of questions to an expert about artificial intelligence, and I want to start quickly. We're talking about AI having a disruptive effect, and everyone is saying something about it. As I just mentioned, you have been thinking about this for a very long time. At this point, it is about the future of artificial intelligence, especially how companies should approach it, how they adapt this concept to their own business. What do you think a company owner, manager, director, people of all levels should consider as a starting point to adapt artificial intelligence to their business? The topic is of course very popular, almost all meetings start with artificial intelligence and end with data. The most important work that needs to be done is to train an AI, use it and be able to carry it out in any way to include it in a scenario. Data inventory should be taken in a healthy way. In general, not all of this data is included in the company. Some data has to be brought in from outside. Integration processes are very critical. Therefore, the biggest obstacle to the successful implementation of an artificial intelligence project will not be the supply of GPUs, as it is thought, but the need. It is primarily the gathering of clean, quality data. Most of the companies also have data. It is stockpiled, archived and unusable due to concerns such as not being brought to the cloud by artificial intelligence. Therefore, the biggest problem is to take this inventory, to determine how it will work in which scenarios, and to produce new scenarios by integrating it with the supply of external data. If we think that a company says short and long term at that time, it is actually short and in the middle, you politely say editing the data, but a little still most of the companies keep the data in excel. There is also such a reality. Data preparation: Is it a departmental setup to prepare for this in the long run? Is it a technology purchase in a business or working with an agency? Is this know-how outsourced? What do you think the best practice here should be? Now, when I make such an assessment in the last year, from 2 and a half to 2,020 to the present, it seems that everyone first thinks that this is a separate field, a longing that requires another expertise, and be a level of education. At the point we have reached, Türkan has turned into a tiny tool that has spread to almost every department. At first, everyone tried to set up artificial intelligence units. Specialized training began to be received on this side, but eventually came. Artificial intelligence has become an element that brings a lot of benefits to democratization. In other words, you need a software developer to solve any problem. As long as the previous topic has ensured the interaction of data and artificial intelligence, business units and end users can now knead technology like a dough in line with their own desires. In summary, this should not be considered as a mere technological leap, but as a sociotechnical evolution. It's not just technology, it's the structure of organizations anymore. Its shape is also beginning to change. Some roles are able to achieve the success of taking. Some roles are disappearing, they can be completely delegated to machines to artificial intelligence. Therefore, at the last point, artificial intelligence has actually turned into a tool that can be used in the hands of every person. Therefore, it is a separate artificial intelligence in companies. I don't think there is a need for a specialized department. There is a lot of such anaphora information on this subject. The department of 60 people was reduced to 10 people. There is a lot of noise about the increase in the work done with artificial intelligence, the fact that people are wasted in this regard, or vice versa, as if a company that did this took people again because it made a loss, of course, I am very curious about your opinion because you have been thinking about it for a very long time, so in fact, the question is, I don't know if I should interpret it, you know, will it cause unemployment? Artificial intelligence can also be it, or artificial intelligence. You can also answer about how the virtual human should develop or transform into their own styles. Let me try to answer both. It is actually very clear that artificial intelligence will create unemployment. This has already begun. Productivity growth is now being announced. You can even make this inference based on the salaries of software developers. For example, almost every application now. Launched the chatbot solution. When there was a need for a new functionality, they immediately started to offer new services without developing any software and almost no software. This paves the way for unemployment. In fact, I evaluate artificial intelligence, especially jenerivia, as follows, it offers you 2 things. It can bring you the information it archives and indexes, which you mostly live in, when you ask about my data, it says that 2,020 is until the end of four, until a date, and it returns to you by saying that it does not have up-to-date data, but if you want, you can search and find the current version on the web. So one is responsible for indexing like this old google did, and the second is the ability to make reznik inference, that is, the coco of the straight man. He has the ability to make all the inferences that he has learned since he was a child, if it rains, he will be greedy, to the point where I need to give you your business registry information and your signature circular in order to give me a credit card. Now, if you are thinking of hiring artificial intelligence as a source of employment and an element of employment, you need to decide in advance which of them you will assign it to. You will ask me what I know. Or will you want your workflow, your business? The first questions and answers were actually done quickly last year, like a rush. You ask questions, we get answers. Now we have arrived. They are called agencies, but this is what it means in Turkish. Factors, by the way, have long been shifting to academie, that is, agent, to another meaning. Ajan is in Turkish, but academia in Turkey has been using the concept of agent since about 1990 years. There is also a lot of controversy about this. You may have caught it in the media, but the academic equivalent of this is actually for this very job, that is, instead of a human being like a human being. Software elements who are responsible for doing a job by exhibiting both social and proactive behavior, in fact, you look at it, so to speak, the software element, when the subject of artificial intelligence is brought here, as a matter of fact, yes, it will have very deep effects that will reduce employment, and it will need to be evaluated sociotechnically and organizational charts will need to be evaluated again and again. Of course, we know that you have been producing on this subject for a very long time in academia, and you have been thinking about it. We know that we were talking to you a year ago, and we were talking about agents, but I want to ask you this, the intersection of generative artificial intelligence and these eygents, and it can be any company. The field of startup can be a tool. What's that intersection point about how you can incorporate that into your own structure? Because everything that is talked about remains very theoretical, and that we translate theory into practice. That's exactly the point. It could be a piece of advice, it could be a point of view, it could be a method, because I know that it's actually business right now. You're thinking about 23 years from now, and it would be great if we could get an idea from you on how to achieve that vision. There is no need to go too far. After 2 years, 3 years, the trend that has actually started to be experienced is that you are getting CRM eygent from a company. You buy another tree for human resources from another company, and suddenly more than one factor involved in the SMEs of the entries and spring-based factors begin to form. He says, and at this point, their synchronization and communication with each other. As a result of this communication, it becomes much more critical for them to behave towards the company's goals. In particular, you can teach the factor you assign to the agonta about its goals, other factors to communicate with, the company's own internal constraints, rules, vision, and mission. Ensuring that they work in harmony with others should be the most important goal. Because when people start to get involved in this business, that is, when people cannot speak the same language as humans, people need to speak the same language as the machine, and this needs to be developed with the culture of the company, the words of the company and the concepts of the company. Therefore, every factor is the intermediary that provides speed service. It doesn't make much sense to take it in and use it. One of the companies that gives an example publishes advertisements on social media during the day when many actions are the subject. They say that they do this proactively by using artificial intelligence factors. However, when considered in terms of the reputation of the brand, that super-intelligent factor that optimizes advertising actually damages the reputation of the company. Because while the agenda of the country is completely different, you want to sell shoes on your e-commerce platform and you are promoting a product to young people that is completely out of the agenda. This is where the story begins. Another factor within the institution is to evaluate the reputation of the brand. It was taken from outside, and that's how the agenda changed. There are such risks and crises in the country and you should manage your brand in this way, he advises. Now these 2 factors can't talk to each other. 2 different departments deal with these factors. On the one hand, you get the warning. The person who receives the warning this month forwards the issue to the person concerned in another department. He has to go and program and manage the other factor. In the meantime, if the advertisement continues to rotate even for an hour when the rust difference occurs, a very destructive environment is created for the brand. At this point, there are few to make the two of them talk. As I said before, the brand's enterprise business, that is, corporate. Managing these factors in a way that complies with the business model, rules and constraints is a completely different issue. You can't buy it from outside either. I mean, is that going to fit in with my culture when you're employing a person just like you? Will he continue to work shoulder to shoulder with me here in 10 years? It's exactly the same as when you're looking for an answer to your question. In fact, if his social intelligence, which has been rumored for years for openaymir, had developed, perhaps he would not have produced the atomic bomb. The concept is a bit aggent based and shir. You talk about the issue of integration into your culture. I want to move on to this a little bit from here. Let's say that as a company culture, the decision was made to invest in artificial intelligence and agencies. But there's a hallucinatory side to my work here. So we have to trust artificial intelligence, yes, we have to increase it to 10 assigned to it, but it also has its own internal problems that can be experienced here. Do you think any manager can trust this business at this point at this time? Whether or not to give a job in a certain vertical to an eygent or still be distant should be approached, in terms of the company's decision-making. Let me try to answer this question through the concept of factor. Can you trust a software, an artificial intelligence, to do a job? In the same way, can you trust a person to do a job? Let's keep the question the same. So let's homogenize the two. How do you trust it? A certain confederates intervel are happening to a person to do a job. That is, on the confidence interval and what we have already envisioned for centuries because it is actually human. Since the factor is a new concept, we have a hard time assigning something to 10. Will 10% answer correctly or 98% answer? It's a bit like it's hard to understand. In exactly the same way, that is, when you are developing an agent solution from a company or yourself, you need to create a test bed environment where you can increase that confidence to 10 by taking the OO isolated document, creating a test bed, creating a test environment and feeding it with various data over a certain period of time. Otherwise, there is not a serious difference between evaluating a friend who will always get the same question and assigning a task to an agent. On the contrary, there is an advantage that you can use another factor to test one factor. In fact, if you control it and a different agency, you go to a factor in the same way. I would like to develop one more factor to question my internal regulations in order to question my legislation. How should the tests for this be? An example for me is 10,000 questions, and he will be able to process these 10,000 questions with questions and answers. Can you perform a simple test bed? When you say that large language models already give you this platform. You should also target the other, you can test it with this. Can you do the same for human? It's more difficult, and it's a different process. I want to ask here and there exactly about the culture, the company culture thing of the business. Let's talk about a company whose culture lasts an average of 20 years. He drank 20 years. In fact, it develops with everything that happens, but the culture we are talking about is a few years old and assuming that we are trying to adapt it, how should we evaluate the cultural adaptation here in terms of the corporate culture of the company on a departmental basis, or what kind of gap training time should be reduced. In fact, there is almost a saying in every word. You know, the best time to plant a tree was 10 years ago. The next best time is right now, he doesn't do it right away. In fact, the answer to this question is that culture, together with natural languages, has actually started to form the basis for following and creating a company's culture, and thanks to this jererity or generative artificial intelligence, it is possible to read and make sense of every text and the company's rules are new. When you write down the constraints and goals it has developed, each of them in natural language, it becomes able to be kept, managed, and interpreted by another inventory in your background. So let's go back to the first problem of companies. So what needs to be done? Whether it is in the data mushroom, whether it is in excels, PDFs, web pages, databases, databases, it is scattered. There are connections between them that humans know about and that AI still doesn't. Productive, artificial intelligence has the ability to make these connections, to keep track of them. The only part of what we call culture is human emotions, you will show 100 laughs to the customer. You will play a team game in your internal relationship, you will empathize. The rest of it can be completely delegated to artificial intelligence. In this, the vereyanin must be created in a healthy way. Even if you didn't, I would go back to the data part. Let's say something like this. This is because the data avante was created from the things that were talked about in the places we met very recently, but as a result, if we assume that the people who created that data are also human and the margin of error, it is built on the mistakes created in the data inventory. In fact, it may as well be a reality. For example, an example was made, the warranty period was given incorrectly by a real employee ageent human in a meeting the other day, and the agent in artificial intelligence gives the wrong result because he learned it from a wrong data. Therefore, when we put so many cultures, emotions, everything on top of each other, wouldn't the error, error, hallucination of the data or the problem or problem that something that may develop here be very big? I think the first phase really becomes more critical than we think. It's definitely contemporary, and it's like he said to comment on the question, how should companies make the transition to artificial intelligence? In other words, 2 types of approaches know that induction can be reached in the solution of almost any problem, especially in the solution of such big problems. In other words, a super artificial intelligence that can know all kinds of details such as planning, documentation, etc., of the entire company that will serve the company. This deduction can be attempted, and one can also arrive at an induction. In other words, a very small customer representative factor can be made below. Very little planning testing factors can be done. In Business, you can decide on your business area according to your business subject. From both of them going from top to bottom to top to top Whatever the fed is, the step-by-step tests here, the confidence intervals, whatever you call it, that big one. Their share will gradually shrink and compare with humans. In other words, if I had already hired someone, you need to squeeze in breaks where you can say that he could have done more than that. You're actually saying that you're going to get better with what you're going to bring up here in time. I want to get into this. We are talking about artificial intelligence, which is developing at an exponential rate, and the subject is not only with technological developments, but also with the technological developments experienced here, ethical regulation, workforce transformation and many parts that are actually reflected in our daily social life. Here, too, we are actually living today. Maybe we can predict the next 12 years, but I'm very curious about what comes to your mind when we think about 5 or 15 years from now, this is the reality we live in. This is a socio-technical problem. In other words, as technology develops, social phenomena change. As social phenomena begin to change, our expectations from technology will begin to change. Right now, when you look at the business lines in general, most of us are dealing with the majority of the population in cities and bureaucracy. Bureaucracy, follow-up, that is, we follow the numbers and data on paper on the computer. As soon as the machines started to take this job, they started right away. What will really be done next? What will that human soul, the human crowd, find creatively, and where will it be headed? It's beyond my capacity. I read Harari Mustafa Suleiman a lot. I've been trying to follow all the authors, but it's really starting to be a very difficult process to predict. Let me give you an example, with the trigger of artificial intelligence, there are big leaps in quantum compiting in the other important field of genetics. Right now, that's specifically about quantum compositing. Let me give you an example, quantum computers would continue to develop at this rate. The production of artificial intelligence models will start to be a matter of time and it is very, very easy. Then we will need a model that has the ability to super-infer on any problem. Really, what are we going to do, software developers? What will happen when the bureaucracy is taken off the field when we can run this job entirely with machines? I can't really foresee it. Considering the approaches of the states to this issue, I would like to ask you about the fact that we have been running a technology company for a long time, because of a few recent discussions, as if we are entering such a cloud of dust, I really wonder about 2000 of them, linux startup software development is actually a concept with the methodology. In the past 15 years, but considering that even half of the code of many technology companies is developed by AI, it is actually about what the publication will provide in terms of product development. At least 10 or 15 years is a lot, but can you give an opinion or opinion on the recent future? The basic hypothesis of the galaxy is that you write code with generative AI, but you don't need to write code anymore. Currently, the analytics of the time spent in the GPT are being published recently. That is, the time is rapidly increasing. The amount of time spent on the web on Google is rapidly decreasing. Therefore, it is clear that all software development platforms from now on will be chat-based. So we don't need a programming language anymore. The machine understands us, and we no longer need corporate screens like we used to. Because the answer to the question we are asking is. There is a platform that can show us the way we want. You ask, how does a company distribute its capital to its shareholders? Normally, how do you expect it in a table or in a histogram in a paychard, that is, you expect a graph and ask for some explanation. You don't need to write any code about it. Ask your question, the necessary answer is either an earthquake from the outside or brought from your internal data sources, and this does not need to be the same image for everyone you like the most. Let it be produced and reflected harmoniously on your screen with the colors you love. That is, in such an environment. The software business is really like this, there used to be a thing, we used to see a lot of magazines saying that software eats the world when I finish the license. Now artificial intelligence is eating software. The word software environment, software platforms, etc., are disappearing. In completely chat-based environments, you will do how a job should be done by chatting. You tell the intelligence, and then it does this by chatting, and what you're talking about actually struck me with a metaphor like this. 40 years ago, the language written in microprocessors and then the language called esem is now completely in daily spoken language as hi level, as if it will open many windows related to software or break doors. In other words, there is a situation like this, you know, the language level was very close to the natural language. Esenbli si c plus plus came obec orand more on languages. Describe a 20 or 25 year process that goes all the way back to Python, and it's always gotten closer and closer to natural language. But at the highest level, there was such a reality. The concepts of the programming language, someone that the machine could understand, were reduced to zero. Now, there are chips in our laptops that understand and interpret natural language. So the machine itself is directly on the silicon chip. He came very close to understanding the bell we were talking about. Is there a need for a screen in this case? Is there a need to encode the images on the screen? Or should the person from this code be software developers? 10 years is not 5 years, I honestly can't see 2 years later. Very interesting. Now I want to get short answers with short questions, because a little bit of the concept of artificial intelligence that we ask this moment, maybe even the question patterns need to change. Quick question, I'm going to ask for quick answers, is there an AI tool that surprises even you? Not at the moment. You have already raised your level of surprise to a lot of the top, because some of these trials in academia are already 5 years ago, 10 years ago, 2010 seven, for example, image processing is most surprising in image processing. We read in academic papers that all the problems in 2,017 image processing were solved. So we were anticipating this. At the moment, for example, I can't say that Google's video-producing and 3 models surprised in any way. In other words, do you think companies should prohibit uploading company data to gang-like tools for the use of their own employees? This is one of the most important flaws, let me answer the question with a question. It's important right now, in many parts of the world there is no short answer, but I'm sorry. I have defended open data for many years, I said that data should be open, but there is a reality like this, imagine that a robot that predicts a magnificent stock has been developed with such a super quantum computer, and thanks to this robot, shares can be bought and sold with very good predictions, and this is the power that controls the robot. Would you consider sharing the data of your country's stock market with this robot? The answer should not be considered, which is such a great value that it is to be able to foresee it. Therefore, it is necessary to open the data of the company and the country to the outside in a controlled way so that it can really use the benefit of this company and this country. But on the other hand, integrations are very critical. In other words, when you only store your data and create layers of firewalls with servers on top of it, you are lagging behind in innovation at this time. That is, with each other of companies. It is also important that they share data and allow new companies to be born. The only thing I will underline here is that it may be new information, but there is a concept called date space, which is the data space that Europe has been working on since 2,010 nine, and as you know, Europe has lagged behind the world in terms of unicorn production. Those who have taken over the European market are always big companies such as American, Apple, Amazon, Chinese Ali Baba. Europe is here so as not to lag behind. Beetho c can't do a successful job either, but beetee be they are just fine. Then he said that our companies should share data with each other. Let's create a secure protocol for this and trigger its innovation in this way. At the same time, when crisis environments such as pandemics occur, I go to my states in favor of my nations and if necessary, go to the data centers in the format I want, with the protocol I want, with the protocol I have previously determined, and come and get that data. He created a building block by saying that I should be able to use it for the peace of my nation and my nations. In fact, in a semi-open way, the data is both open to the outside and completely controlled, accessible. We hear about a similar protocol in Turkey under the name of public data space, but I am really curious about its output. Every country can be in company partnerships, it can be within the company. I think we should follow closely about these data fields. Right here, I'm actually going to ask. You can evaluate the approach to the question in Turkey in the world. Do you find the laws and regulations on artificial intelligence sufficient in the world? This is actually a vicious circle as the regulation is issued. You need more artificial intelligence because it is to train people to read, understand, interpret and act on them. It is very difficult to manage them. It turns into an egg-and-chicken relationship. In other words, is it really necessary to work under control with prohibitions and sanctions here, or will the bureaucracy completely eliminate the subordinate. Is this paving the way for artificial intelligence? I hope that our elders will make the right decision on this issue, and then I come to this, do you think countries should establish artificial intelligence ministries? The first problem is actually do you think companies should establish an artificial intelligence department? I think it should be democratic and accessible to everyone. First of all, I think the Ministry should be protected. Or the data institute has a unit called Detay You K in the UK. So even if you are going to buy a remote control on any television in the UK, its protocols are defined in the post and the data you k. You can find it listed. In other words, the standard of the data to be used by all data is predetermined by protocols. Therefore, what can be done with this data after we first draw up the data map of Turkey? We need to think about this with artificial intelligence. Because only as soon as we say let's develop Turkish hands, that is, the elders have already done it, understand the natural language, etc., the important thing is that Turkey can solve the problems here. I go back to the inference part, the lines of business that can make inferences will be followed. We need to develop special smaller language models, small and large models. My last 2 questions are a little more off-topic here. Is there any TV series or movie that you think works best with artificial intelligence? You liked watching it a lot, or I can recommend a book rather than a movie that deals with the concept of artificial intelligence very well. Or let me say author, suggest author. Right now, this work is artificial intelligence, in fact, it's not really smart right now, it's very large statistical models that predict human footprints. The main thing is consciousness. There's a big gap in terms of when this machine is going to be conscious, or if it's really conscious right now, how we're going to approach it to 10. It's very good. There are 2 operative authors. The father of science fiction is Saydam Ayzek, Asimo This is definitely a must read in the foundation series, and there is even a special universe term in his foundation series. We named the galaxy after the company. Another is Sanisila Russian writer. Even if you read the short stories of these two. In almost all of the current science fiction movies, I don't want to talk about it too much, but in all of the ones I've seen, it's a trace. In all of them, the scenarios there have already been processed. Let me tell you that both authors will give you a horrible imagination. Super advice then speaking of consciousness, I come to the last question. Could artificial intelligence one day take control of the world or the universe? Let me answer the question with a question. If he takes over, will we know about it? Then maybe he took over and that's really the point. I have 5 family stories that rule the world and so on, maybe it's true, maybe it's getting used to it, but I think the human organization is self-organized, self-adapted, the concepts that are used a lot in academia, that is, we are like a flock of birds, we fly somewhere together. Sometimes, once in a while, pioneers come out and change our direction, but I think the environment is completely self-organized and self-organized. Now some are among us. There are robot birds. Do they rule us or not? I think let's continue to see what would happen if they managed and what would happen if they didn't. I don't think there's much to do about it. Elon Musk is doing the right thing. We have to leave this planet and find an alternative planet. The issue goes in that direction. So erdem thank you very much for sharing an idea that you participated in, today we talked about what awaits us in the coming years to grow technology, erdem eser second. If you want to be informed and support our new videos, it's up to you to subscribe, like, comment or share so much for having me to see you. Thank you, a love regards.