"Fancy documentation pages with flashy color palettes and animations are a thing of the past (Tired). Instead, a simple single Markdown (.md) file and a single 'copy to clipboard' button are the new normal (Wired)." The reason I resonated deeply with Kapasi’s writing recently is simple: the primary target audience for the products, services, and libraries we create is no longer people, but artificial intelligence. The Coming of the LLM Era: The Rise of Machine-Friendly Interfaces As anyone who has recently used the collaborative work function ('Work with') provided by OpenAI's MCP (Model Context Protocol) or ChatGPT's desktop app will have noticed, the UI is much more effective when it is easy for machines to understand rather than people, that is, LLMs. This is a trend that redefines the traditional concept of User Experience (UX). For a long time, we have been creating UIs optimized for people with beautiful interfaces and friendly explanations. But now, things are different. More and more things are being done by LLMs or agents instead of people. Why does LLM hate UI for humans? It's clear why LLMs shy away from complex and flashy interfaces for people. LLM wants to scrape, not explore. It needs simple, clear text structures rather than complex menus or navigation. LLM prefers text over visual stimuli. Information is presented more effectively in concise text than through animations or videos. LLM likes Curls, not Clicks. It prefers direct access to data provided in API form rather than following links and clicking. These characteristics become increasingly important as LLMs become more and more commercialised in a wider range of fields and become deeply embedded in our daily lives. New challenges for product developers and service providers It is now time for developers and service providers to make their products and services LLM-friendly, which requires a fundamental change in the UI. All features should be expressed as concisely as possible in text-based terms. Data must be presented clearly, either as an API or in minimal documentation format. You should actively utilize simple formats such as Markdown to organize information in a structure that is easy for LLMs to understand. Back to the age of text? Technological advancements always happen faster than expected, and this LLM-centric interface change will also come to us quickly. Considering that protocols like MCP and LLM-friendly functions are becoming more and more standard, it is necessary to reorganize the way products and services are provided even now. I wrote the above article a while ago, and it seems to be approaching quickly. If your product is going to survive into the future, it needs to transform into something that 'LLMs love'. It needs to be simple, efficient, and easily accessible. Take a look at your product, service, or library now and think about it. "Now, if my main customer is LLM (AI), how should this function change?"

Recently, Twitter (X) founder Jack Dorsey and Tesla CEO Elon Musk have been making headlines by calling for the abolition of intellectual property (IP) . When Dorsey made the radical suggestion that “all IP laws should be abolished,” Musk simply responded, “I agree.” Since both are considered icons of innovation, their comments caused quite a stir. Their claims are too significant to be dismissed as mere “flashy remarks.” What is the real intention behind the argument for relaxing/abolishing IP laws? Jack Dorsey and Elon Musk have a clear reason for calling for the abolition of the intellectual property system. They criticize that the IP system does not promote creativity and innovation, but rather encourages monopolies and abuse of rights, thereby denying creators fair compensation. Musk has previously said, “Patents are for the weak,” and has disclosed Tesla’s electric car patents. Jack Dorsey also supports open source projects like Blue Sky, believing that the openness and sharing of technology leads to innovation. Their argument ultimately raises the following question: "Is it really in the public interest to continue to recognize the right to monopolize knowledge and technology?" Democratization of Information and Knowledge: A Debate That Has Existed in the Past In fact, this is not the first time this discussion has taken place. There have been cases in the past that showed how explosive innovation and growth can occur when knowledge and technology are not monopolized by a certain class but are open to everyone. Search Engine (Google) : Google, which collected and provided a large amount of content on the web without permission, initially had a controversy over copyright issues, but eventually opened the way for mankind to obtain vast amounts of information immediately thanks to the concept of "Fair Use." Recently, the Book3 project and various media companies are also in copyright lawsuits with AI companies due to similar lawsuits. Wi-Fi technology : The government opened up the wireless frequency so that anyone could use it, and the technology standards were made open source, so now we can connect to the Internet anywhere. When you were young, you know the fear of pressing the center button on your phone by mistake and getting a huge bill? That fear has disappeared these days. When I think about it now, I realize how unfair it was. Open source software (Linux, Apache, etc.) : Let's make the source code public and share it freely. Despite initial criticism, it eventually led to innovation and development in the global software industry. All of these cases show the lesson that innovation can occur more quickly only when knowledge and technology are open . On the other hand, there are historical cases where excessive monopoly rights have hindered technological progress, such as the 'red flag laws' that hindered automobile innovation in 19th century Britain or the Wright brothers' patent wars in the early 20th century. But how will we solve the problem of creator protection? The argument for abolishing intellectual property rights is attractive, but there are also many practical concerns. Without IP protection, creators and innovative companies may not receive fair compensation, which may reduce their motivation to create and invest. The arts, in particular, are already suffering from the problem of AI absorbing creative works without permission. In fact, what they are saying is not to eliminate copyright altogether, but to allow free use and provide proper compensation. Musk and Jack Dorsey did not say, "Let's just get rid of intellectual property rights!" but rather that the relevant laws are too much of a hindrance to technological development. Copyright law relaxation/abolitionists like them propose the following methods. Compulsory License : A method that allows the use of copyrighted works without the consent of the rights holder if certain conditions are met, but requires compensation at a set rate to be paid to the copyright holder. Revenue sharing model : This is a method of sharing a portion of the revenue generated by the AI model with the data provider (creator). In fact, some companies are already adopting this method. Shortening and flexibilizing the term of IP protection : This is a way to make knowledge available to the public domain more quickly by shortening the term of IP protection. Expanding government and public support : This is a plan to have the government or public institutions compensate for creative activities themselves, thereby releasing creative works into the public domain. These compromises are realistic approaches that ensure fair compensation for creators while allowing more people to access knowledge and technology. In fact, some of them are being applied on YouTube, TikTok, etc., such as allowing the original creator to receive revenue even if certain music is used without permission, or requiring unauthorized users to notify viewers and limit revenue. Where will intellectual property rights go in the AI era? The future of the intellectual property system ultimately comes down to this question.

After writing the above post, I received many questions from people. People who are executives or managers, not individuals, or who are running a one-man business and doing various things at the same time. In the early days of technology, it was common for organizations to have a clear choice between developing software themselves (in-house development) or developing it through an external development company (SI). This was a relatively simple choice structure, but it often caused confusion and difficulties in the development management and maintenance process. Recently, as the technology environment centered on platforms and clouds has developed, the possibility of more flexible linking and utilization of different technologies and services has opened up. However, this free combination method poses new challenges to the integration and management of each technology, and requires more meticulous coordination and management capabilities than before. Division Before AI Introduction After AI Introduction Marketing Operations Production/analysis, strategy/creative and other roles are separated and operated independently. The boundaries between IT and the business are blurring, and the scope is becoming broader and more complex. Strategy and Creative There are many restrictions, and the pace of experimentation and trials is slow. AI removes constraints and enables faster, broader creative experimentation Key Message Each role is clearly divided and limited operations are focused on efficiency. AI makes entire organizations more agile and accelerates creative strategy execution In this process, the technology barrier is gradually lowering. In the past, IT professionals exclusively managed technology tools, but now, thanks to the advancement of low-code, no-code, and AI-based automation technologies, technology can be easily utilized by non-experts. Now, the ability to utilize technology is becoming an essential competency across the entire organization, not just the IT department. The key concept here is Martec's Law. Martec's Law is a law that states that technological change progresses exponentially, while organizational change progresses very slowly. Organizations should be aware of this gap and selectively adopt technologies in line with the organization's strategy and goals, rather than blindly accepting all technologies, and increase agility. Now, the AI technology environment is not just a tool for increasing efficiency, but a foundation for creative strategies and experiments. Organizations must recognize this new possibility and provide members with sufficient autonomy and opportunities for experimentation so that they can implement creative and innovative strategies without being too immersed in technology. In addition, technology-driven operations are increasingly becoming more complex and unpredictable. The ability to orchestrate digital operations is essential to harmoniously operate various AI-based agents and tools. Just as a conductor tunes various instruments, the ability to efficiently integrate various technologies and AI tools to meet the organization’s goals is required. Now planning and execution will happen simultaneously. Shift from the early, clear technology choices (build vs. buy) to platforms and hybrid models

Since I started blogging in the early 2010s, I have constantly been thinking about how to monetize my content. At first, I made money by writing advertisements or sponsored articles through Google AdSense or blog experience groups. However, after I started working, I started to focus more on the records themselves rather than on the profits. However, around 2020, the number of newsletters exploded, especially overseas, and an atmosphere of “willing to pay for good articles” was formed. I followed that trend and published a paid newsletter, gaining tens of thousands of subscribers, and in the process, I even had the experience of publishing a book. Of course, it wasn’t easy to make a huge profit as much as I put in the time and effort. Nevertheless, I continued to write because I liked writing and recording my experiences. This trend is well known overseas as Substack . When you first post, there is no special cost, and when a paid subscription occurs, the platform takes a portion of the subscription fee (about 10%). This is a structure that is not burdensome when you start, but has the disadvantage that the 10% commission becomes significant when the number of subscribers increases significantly. That is why the 'flat rate model' that pays a monthly fee, such as Ghost or Beehiiv, is popular recently. Since the amount to be paid to the platform is fixed or increases only in stages even when the number of subscribers increases, it has the advantage of "the more profit, the more I take." In fact, you can find similar trends in Korea. For example, there are newsletter publishing services such as Stibee and Mailly , and the fee system varies depending on the size of the sending, and in the case of paid newsletters, a separate fee is charged. Since each platform has slightly different functions, templates, and reader analysis tools, it is a good idea to look at the items below when deciding which service to choose. First, the ‘size’ and ‘expected number of readers’ of the newsletter (or blog) I will publish. Second, how technically involved do I want to be in the setup and how much convenience do I want to be provided? Third, does the platform fit with my values or operating philosophy (political/social issues, community policies, etc.)? In fact, no matter what newsletter or paid content platform you choose, I think the power of the content itself is ultimately the most important. In my case, as I wrote and met readers, my personal brand was naturally formed, and I met people who supported me for a long time. These people are the driving force that allows me to continue writing. Also, even if I didn't make a lot of money at first, I felt rewarded enough just by the satisfaction I got from recording the record itself, and when I looked back later, I felt like I had left behind these experiences. Of course, in order to systematically increase profits, you have to carefully consider economic aspects such as platform commission rates and monthly subscriptions. Looking at overseas cases, there are writers who say that their profits improved when they moved from Substack to other platforms (Ghost, Behive, etc.) even though the number of subscribers did not increase significantly. Since a 10% continuous commission is not easy, it may be more profitable to pay a certain amount and take all the remaining profits. Platform Pricing Structure Types Is there a free plan available? Paid Plan Starting Price Fee Policy Behive Flat rate (Based on number of subscribers) Yes $49/month (Less than 1,000 people) No fees Ghost Flat rate (Based on number of subscribers)

This is a simple and clear process that goes from problem recognition to goal setting and implementation based on the three key questions (WHY, WHAT, HOW) that are often mentioned when establishing a “Strategy.” It’s good that I’ve organized it like this instead of just talking about it, and it’s become clear to me. This is something I often talk about at the beginning of a workshop program, so I thought it would be good to share it. The first question that comes to mind when thinking about strategy When you say “Strategy,” many people think of a grand and complicated process. There are various frameworks, such as management scholar Michael Porter’s five competitive factors model, value chain analysis, and the BCG matrix. However, overly complicated tools and theories can actually reduce execution ability or cause important issues to be missed. Ultimately, a good strategy depends on how clearly you define the three questions of “Why is change necessary (WHY), what should we go for (WHAT), and what actions should we take to achieve that goal (HOW)” and how you prepare specific implementation plans. In some ways, this seems simple, but in fact, it is the essence and core. ⸻ WHY change? Why change? In the current situation, why change? The starting point for strategy development is problem recognition. You need to look at the current situation (“Current”) of your organization or team and start with the questions, “What exactly is the problem?” and “Why is this problem important?” A strategy without consensus on the 'problem' is dangerous. For example, if there are members who think that “it’s not a big deal” even though the market share is falling, it is difficult for the organization to establish a strategy. “Why is this problem serious?” must be defined first, and only then will all members agree on the “need for change.” Example: If a startup experiences a sharp increase in the churn rate of its app users, the severity of the problem should be shared by all internal members by specifically explaining the potential risks and financial losses that this churn rate would cause. Criteria and sources for defining the problem The important thing here is that the problem must be verified through data or objective indicators. It is easier to build consensus and a sense of urgency among members when it is based on ‘facts’ rather than ‘feelings.’ Source examples: recent customer interviews, user experience (UX) research, market research, financial reports, etc. Statistics: For example, if there are statistics that say that more than 90% of startups worldwide will fail within 5 years in 2023 (source: CB Insights report, etc.), the problem can become clearer under the premise that “we cannot be an exception.” WHAT is winning? What does the 'future' we want to achieve look like? The second is the goal setting stage. If the “problem we need to solve” has become clear through WHY, now we need to draw a concrete picture of “what kind of future is the ideal state (Winning) that we think of?” This determines the organization’s vision (Ambition), direction, and a kind of “North Star.” Set visible and specific goals You may be thinking, “Isn’t a vision that is too big and abstract useless?” However, setting a specific vision is important because it provides a reference point for decision-making and prioritization. Example: “Achieve 30% domestic market share within 2 years” has specific numbers and time frames. This makes it clear which projects to allocate resources to first and which features to prioritize. Presenting a vision of the future and building consensus A vision cannot simply be “written down in an internal document and left behind.” It must be something that members can feel the value of the goal and be convinced of, “Is this really achievable?”

I recently met a friend who is actively involved in environmental social movements. She, like Thunberg, says that we are destroying the Earth just by using search engines and that it has various effects on the environment. From carbon footprints to planting trees, it is a topic that has been frequently heard since the 2010s, so I was indifferent, but I didn't expect it to extend to artificial intelligence... Should I say it is natural? Well, let's talk about it one by one. Artificial Intelligence, IT Services Destroy the Earth? Recently, there has been a flood of warnings about the environmental impact of generative AI services, especially chatbots like ChatGPT. However, many of the claims are somewhat exaggerated or delivered in a sensational way without context. Of course, it is true that environmental costs occur in data center operations or in the process of AI training and learning, and the claims of environmental groups cannot be said to be completely wrong. However, when looking at actual data, we can see that the claims are greatly exaggerated. Let's examine the truth together based on accurate data. ChatGPT consumes 10x more power than Google Search? To conclude, “It’s true but it’s wrong.” In other words, the ratio itself is roughly correct, but the actual consumption is greatly exaggerated. The claim that “a single chatbot query (2.9 Wh) consumes 10 times more power than a Google search (0.3 Wh)” is actually based on old data released by Google in 2009 and a semi-analysis from early 2023. According to the latest research as of 2024, the actual power consumption of a single Google search has decreased significantly to about 0.04 Wh, and the power consumption of a typical chatbot question is also about 0.2 Wh. In the end, chatbots still consume about 5 times more energy than Google searches, but the absolute power consumption itself is very small. Big tech companies need to improve power efficiency, if only to increase their profits. They are still thinking about how to use the power they use efficiently. In other words, there is a slight misconception that generative AI (such as ChatGPT) consumes more energy than Google searches and that this will continue. To put it simply, if you use the chatbot 100 times a day, it will consume about 20Wh, which is equivalent to driving 30 meters. In other words, if you use it for a year, it is equivalent to driving back and forth to a nearby restaurant. “AI usage” consumes a huge amount of electricity? This is also an exaggeration. Even if a typical user interacts with a GPT-4-level chatbot 100 times a day (about 7.2 kWh per year), this is similar to or less than the energy consumed by driving a gasoline car for 10 km (about 7.6 kWh) or taking a few warm showers. In other words, chatbot use itself is not an environmental disaster, but rather an energy usage that can be easily saved in daily life. Rather than worrying about AI usage, it is much more effective to save in other lifestyle habits, such as walking short distances or reducing shower times. I took this opportunity to look at the presentations of those concerned about the environment, such as the UN and Greenpeace, and it seems that they are writing without distinguishing between AI training/learning/research and use. The infrastructure operated for service use is significantly smaller than the infrastructure used for training, etc., and as mentioned above, it is a small amount compared to the amount of electricity consumed in daily life. Is AI really a water-drinking hippopotamus? The claim that “ChatGPT evaporates a bottle of water (500 ml) in a few questions” is also a misrepresentation of the facts. The original study said that it uses about 0.5 L per 20 to 50 questions, which includes not only the cooling water used directly in the data center, but also the water used in the process of generating electricity and the water used during AI training. In fact, the amount of cooling water used directly in a data center is only about 0.5 L per 300 questions (about 30 mL per question). This is a small amount compared to the amount used for downloading an app (40 mL), streaming music for an hour (250 mL), or a Zoom call for an hour (1720 mL). More importantly, most online activities use similar or more water. The misconception that AI services are particularly “water-wasteful” is far from reality. Summation Environmental groups’ claims that data center operations and AI training incur environmental costs are not entirely wrong. However, AI use itself is not an environmental disaster that will destroy the planet, and most of the claims are exaggerated. AI data centers may put a strain on power and water resources in certain areas, but this is a common issue with all Internet-based technologies. AI’s energy efficiency is constantly improving, and its environmental burden is still minimal compared to other activities. The real issue we should be concerned about is not refraining from using AI, but rather shifting our energy production to renewable energy, focusing on designing data centers that are water efficient, and reducing unnecessary energy waste in our daily lives. The environmental costs of technology are clear, but understanding the context and scale, and pursuing more fundamental environmental changes, is the real solution. I understand the urgency of environmental groups, but I hope they don't exaggerate the situation or misrepresent the information.

We often experience unpleasant things in our lives. For example, when a colleague at work doesn't respond to an email I sent, when a friend completely forgets an appointment and doesn't show up, or when a restaurant employee makes a wrong order or acts rudely. When this happens, many people instinctively think, "Does that person have ill feelings toward me?" or "Is he ignoring me?" I have had many experiences like that. If you start to think about each and every action of the other person and think, “What did I do wrong?” or “What kind of feelings does this person have toward me?”, you will become mentally exhausted and your relationship will become uncomfortable. Then, I came across a sentence in a collection of English-American proverbs. “Never attribute to malice that which can adequately explained by incompetence.” If something that can be explained by malice can also be explained by incompetence or mistake, Don't take it as malice, but as a mistake or a lack of ability. When I first saw this phrase, I was struck by a strange shock and had a profound realization. After looking into it a little more, I found that a similar expression also appeared in the German writer Goethe's "The Sorrows of Young Werther." “Misunderstandings and neglect create more confusion in this world than trickery and malice.” Misunderstanding and carelessness cause more chaos in this world than malice and cunning. These two phrases are expressed differently, but the essence is the same: most of the unpleasant experiences we encounter in our daily lives are not due to malicious intent on the part of others, but are simply due to mistakes, carelessness, or lack of ability. In fact, many studies and psychologists' opinions say that when evaluating other people's behavior, people often interpret it based on their own emotions or preconceptions rather than their actual intentions. In psychology, this is called the 'fundamental attribution error'. When seeing other people's negative behavior, we tend to ignore the situation or environmental factors and attribute it to the person's personality or intentions. If this trend continues, we will live in unnecessary stress and misunderstanding. Even in the workplace, if you see a teammate's mistake and think, 'They are intentionally trying to make me uncomfortable,' teamwork will collapse and it will be more difficult to solve the problem. However, if you understand the mistake and help each other, the relationship can move in a more positive direction. When I applied this principle to my life, my relationships became much more comfortable. When I don't get a reply to an email, I think, "He must be busy," or "He must have forgotten," and give him a light reminder. When a friend is late or can't come to an appointment, I think, "He must be busy or he may have forgotten," and I don't have to blame him or feel hurt for no reason. Of course, there are people in reality who are truly malicious. However, most problems in our daily lives are not explained by malicious intent, but rather by simple carelessness or mistakes. Just remembering this can save us from considerable stress and unnecessary conflict. The next time someone's behavior makes you uncomfortable, why not think about this principle? Of course, there are times when you can't stand it, but what can you do then? You have to blow it up.

Have you ever heard the story that “ChatGPT could replace Google Search”? Just a few months ago, there were concerns that Google was in danger of being overtaken by ‘AI chatbots’. However, Google has now established a comprehensive AI lineup with a new project called ‘Gemini’ and is upgrading its services one by one. In this article, I will look at Google's actions since the introduction of ChatGPT, and the background of the ' Gemini Project' created by DeepMind and Google Brain . I will also answer the question, "Can Google really change the AI landscape?" There were times when it was dangerous and there were also unexpected mistakes, but I will talk about Google's unique tenacity and know-how that emerged in the process. Please look forward to it! Where did Google go?”: The crisis that emerged after the release of GPT-4 Since late 2022, OpenAI's ChatGPT has taken the world by storm. It has attracted more than a million users in just a few days since its launch, and people have been so enthusiastic that they are even saying, "I'm using ChatGPT instead of search." This was a serious warning to Google. Google Search has been the company's core source of revenue and the "gateway to the internet," and the news that a chatbot could replace it shook the market. Limitations surrounding LaMDA: In fact, Google had an excellent language model called LaMDA internally. However, due to its overly cautious attitude, it greatly limited the scope of its disclosure, and the demonstration was limited to a 'dog story', so it did not make its presence known. Wall Street and Internal Unrest: Google, which had been touting its “AI First” motto, showed a passive attitude toward the popularization of AI due to ChatGPT, and Wall Street showed unrest, and Google’s stock price fell nearly 39% year-over-year. Researchers also left the company or became exhausted without any results. So CEO Sundar Pichai and the management team took extraordinary measures. They gave the order to “come up with a product that can compete with ChatGPT within 100 days.” This task was assigned to Sissie Hsiao, a Google veteran who had worked at the company for over 16 years. Google's AI Solution: The Launch of Bard and Its Ups and Downs Shishi Xiao and technical lead Eli Collins had to build a chatbot in a very short period of time, and they used the internal project name “Big Bard.” However, the initial Bard was developed based on a smaller ‘PaLM model’ rather than LaMDA, which meant that they focused on delivering results as ‘quickly’ as possible by investing time and resources. • Mistakes immediately after release: In February 2023, Bard's public demo was criticized for providing incorrect information regarding the 'first observation of the James Webb Space Telescope (JWST)'. This incident was reported in the media, and Google's stock price fell by 7% at one point, causing negative effects. •Internal backlash: Some employees were dissatisfied with Bard, going so far as to call him a “sick parrot.” There were also instances of biased responses, such as racial and gender stereotypes, and vulnerabilities in the accuracy of the content. In the end, Google released Bard as an ‘experimental’ product, limited to certain regions such as the US and UK, and began large-scale training. It was the first AI chatbot from Google in a long time, but it seemed like it still had a long way to go before it could meet the expectations of both inside and outside the company. 'DeepMind + Google Brain' Integration → New Breakthrough, Gemini Project At this point, Google made a decision. It decided to merge DeepMind, based in the UK, and Google Brain, based in the US, to form a more powerful AI research organization. The news that DeepMind, which created the famous AlphaGo, and Google Brain, which can be said to be the birthplace of the Transformer architecture, were meeting created a buzz in the industry. Integration Background: The main purpose was to solve the problem of ‘lack of consistency in AI strategy’ internally and prevent talent drain. A consensus was formed that the two organizations should cooperate under a clear goal rather than operate separately. Leadership changes: DeepMind CEO Demis Hassabis will lead the combined organization, while Google Brain’s Jeff Dean has moved to the role of chief scientist. Internal friction: However, Google Brain staff were unhappy with the integration that would put DeepMind in the lead, and some even moved to competitors such as OpenAI. As a result, the integrated AI organization that was born this way made the “Gemini Project” its top priority. And the huge task of combining Bard and Gemini, led by Shishi Xiao, and further embedding AI functions into all of Google’s services (Search, Gmail, Docs, YouTube, etc.) began. Gemini 1 finally revealed: its light and shadow In December 2023, Google finally released Gemini 1, which was evaluated as “as good as GPT-4.” Sundar Pichai also said, “This is one of the biggest technology launches in Google’s history,” so the internal expectations were high. From this point on, Google rebranded Bard as Gemini and declared, “We will show you a new paradigm in AI.” But the world of AI isn't always that simple. Image generator controversy: In early 2024, Gemini's "image generator" feature faced user backlash for overly racially diverse images (e.g., historical figures were always depicted as being of different races). AI Ethics vs. DEI (Diversity, Equity, and Inclusion): Within Google, there were instances where the ‘realism’ of AI results was distorted while trying to adhere to DEI policies, and this led to the company suspending the image generation function for a while.

This is by far the question I hear most often these days. Regardless of their position or role, whether they are developers, designers, planners, marketers, or CEOs, whenever I meet them recently, they all bring up the topic, “AI is developing so rapidly, so what should I do to prepare?” “New technologies keep coming out, so I feel like I need to keep up, but I don’t know what to do,” or “I feel like I need to do something, but I’m afraid I’ll fall behind, so I don’t want to be too scared.” Of course, this kind of scenery is not so new anymore. Even now, in 2025, after the era of ‘generative AI’ that exploded in earnest in 2023, people’s curiosity is still the same. How can artificial intelligence be applied to my work, and what should I study or prepare for? 2025, more diverse workplaces and people Personally, my work has become much more diverse since 2025. I deliver products to public companies and provide technical consulting, teach machine learning (ML) at universities, and travel around the country to give special lectures on AI. As a result, I get to meet a wide range of people. People of all ages, majors, and industries, and in academia, people from undergraduates to PhD students are trying to learn and utilize AI at various levels. When I meet and talk with people from various fields like this, I feel common concerns and desires. It's the point of " AI is so important, but what exactly should I do?" Some people have no programming knowledge at all and feel lost, and some people are already good at programming but worry about 'should I study models or start with data science?' And marketers, planners, and executives are most curious about "Where exactly can our company apply AI to achieve results?" Smartphone proliferation vs. AI proliferation I got my first smartphone (iPhone 3GS) in 2010. Before that, I didn't really use a cell phone in high school because my parents thought that "cell phones are something that distract you from your studies." But right after I entered college, a new wave of smartphones was just starting, and I was able to jump in at the right time. At that time, smartphones literally changed the world. The “mobile era” began around 2010, and people quickly adapted to devices and apps. However, looking back, back then, even if the device itself was different, the basic functions that could be performed were quite clear. Just taking pictures, making phone calls, using messengers, and uploading to SNS were all good enough to enjoy a high level of convenience and value. However, the era of artificial intelligence is a little different. AI is not a technology that can provide the same utility to everyone just by giving them the same 'device'. For example, even if you have the latest large-scale language model (LLM) or high-performance machine learning infrastructure, if you do not have the ability to handle it properly, it will only be limited to functions such as "asking about the weather" or "summarizing articles." From the perspective of consuming AI services, simply 'listing functions' will not be very attractive. Where does the “AI gap” come from? Infrastructure vs. Utilization Capabilities Smartphones have become devices that everyone can hold in their hands, so 'distribution' itself has become the key. However, AI is a technology that requires huge infrastructure such as data centers and cloud GPUs, and it is not easy for ordinary individuals to directly own and operate them. Instead, cloud services have developed so well these days that you can rent AI models or GPU resources just by paying money. The problem is "how to utilize them." In the smartphone era, no matter how many new features there were, it was convenient enough to just use the basic functions like 'camera' and 'calls' well. However, in the case of AI, the impact may not be that great if only the basic functions are used well. Diversification of User Needs As services utilizing AI become more diverse, the functions and goals desired by each person are clearly different. Some people want to shorten the design sketch work by utilizing image generation AI, while others are looking for ways to efficiently summarize and extract large Excel documents through data analysis AI. We are entering an era where the uniform approach of “solving all problems with one powerful model” is no longer effective. As services utilizing AI become more diverse, the functions and goals desired by each person are clearly different. Some people want to shorten the design sketch work by using image generation AI, while others are looking for ways to efficiently summarize and extract large Excel documents through data analysis AI. We are entering an era where the uniform approach of “solving all problems with one powerful model” is no longer effective. ‘Individual capabilities’ have become more important Looking at this situation, I think that in the future AI era, the gap will be based on 'who can use AI more creatively and effectively' rather than 'who has a more powerful device (or model)' . In the past, you could enjoy a certain 'advantage' by just buying a more expensive smartphone and having a faster internet environment, but now the axis of that advantage is changing. “AI Agent” and MCP (Model Context Protocol) The concept of AI agents and MCP, which were hot topics in 2023, will be discussed even more actively in 2025. Simply put, it means that the role of humans connecting various 'tools' to artificial intelligence and directing and supervising how to use those tools to solve problems will become important. It is not perfect, but it could take the form below. It is a concept that is a bit more advanced than the concepts of AEO/GEO that existed until recently, but in the end, if the structure and data were made easy for people to see and read until now... Now, it is a concept that makes it easy for machines to read and find. For example, by connecting additional plugins, APIs, etc. to a chatbot or LLM model, data can be retrieved, documents can be read, images can be created, and these can be synthesized to obtain meaningful results. At this time, the ability of the person handling the agent directly determines the quality of the results. The key is not simply a question like “Tell me the weather,” but how well the complex request, such as “Perform this analysis based on this data, and automatically create visual materials based on the results,” is designed. A new direction for UI/UX

If you look at history, the greatest innovations in technological advancement have always started from 'simplicity'. Just as USB integrated numerous complex dedicated ports into one, a similar change is happening in the AI field now. At the center of it all is MCP (Model Context Protocol) . MCP, 'USB-C Port' in the AI Era MCP is a standardized protocol that connects AI models with external data sources and tools. Previously, if AI wanted to connect to various data sources such as Gmail, Slack, Google Drive, and internal databases, it was necessary to develop separate integration methods for each API. This process was complicated and cumbersome, like managing multiple doors with different keys. MCP is the “USB-C port” of AI that solves this complexity in one fell swoop, allowing you to connect numerous tools and data sources with just one standard protocol. MCP vs. Traditional API: The Revolutionary Difference Traditional API approaches require different authentication and integration methods for each data source, like creating a key for each lock. MCP changes all that. Standardizing the integration approach using a single protocol Supports real-time two-way communication to synchronize data in real time Dynamic tool discovery allows AI to automatically identify available tools Adding new tools is very simple with a 'plug and play' approach Provides consistent security control capabilities MCP Architecture Made Easy Just as you can connect multiple devices at the same time by plugging in different cables to a USB hub connected to your laptop, the MCP client makes it easy for AI to pull in data from a variety of sources. The MCP architecture consists of: MCP Hosts : AI applications that access external tools or data, such as Claude Desktop and ChatGPT. MCP Client : Software that requests data required by AI and connects to the MCP server. MCP Server : A server that exposes specific data sources, such as Gmail, Slack, or a local file system, through MCP. Local data sources (Local) and remote services (Remote) : Data repositories and external services that the MCP server actually accesses. The Power of MCP Seen in Real Business Cases The value of MCP is further demonstrated through real-world usage scenarios. Travel Planning Assistant : Check calendar schedules, send emails, book flights, etc. in one MCP for real-time management

Recently, there have been a series of political declarations that “we will become an AI powerhouse.” However, in reality, there are sighs on the ground that “it should not just be a slogan, there is too much infrastructure that needs to be taken care of right now.” In reality, our power situation and energy infrastructure are still reaching their limits. Today, I would like to talk about data center distribution and local power infrastructure. In fact, the power issue has been an issue that has been consistently pointed out for several years. It seems that it has suddenly emerged as a task that can no longer be delayed, as it is intertwined with ‘artificial intelligence (AI)’, which has recently become a hot topic. Power shortage in metropolitan area, power surplus in local areas? First of all, looking at the statistics for the past several years, the metropolitan area is already in a state of power shortage, while the provinces, on the contrary, are in a state of abnormality where there is power surplus. According to Korea Electric Power Corporation statistics, Seoul’s power self-sufficiency rate is only 8.9% as of 2022. However, in provinces such as Chungnam (214.5%), Gangwon (195.5%), and Busan (216.7%), there is power surplus at an excessive level. In addition, as the AI industry becomes more active in the future, various data centers and research facilities will be built, and these facilities will consume a huge amount of electricity. Several major companies' IDCs (Internet Data Centers) and universities and research institutes are already concentrated in the metropolitan area, and there are already signs that the metropolitan area's power is nearing its limit. Division 2017 2018 2019 2020 2021 1st place Seoul National University Seoul National University Seoul National University Seoul National University Seoul National University 2nd place KT Mokdong IDC KT Mokdong IDC LG Gasan IDC LG Science Park (EAST)

AT fields in , area expansion in , and unique barriers in the . Those who encounter these for the first time might usually just dismiss them as “cool fantasy settings.” However, if you think about it a little, you will realize that each of them represents an absolute barrier to protect “one’s own world (self).” In , the AT fields symbolize the “wall of the heart,” and area expansion and unique barriers also implement the user’s personal beliefs and values into extreme spaces. Therefore, these devices metaphorically show situations where communication and rapport are difficult, or “division.” And this metaphor is not just a story in anime or games. In our daily digital environment and in the political and social fields, we witness deep “division” in the way each person firmly protects their own barrier. When I first saw Evangelion as a child, I was very shocked. The unique characters, the production, and the deeper the content, the more I learned about it... By the end, it was a work that made me think about understanding, love, and harmony, and further, it made me think about totalitarianism and individualism. In the end, Evangelion just unites people in front of the grand name of the "Humanity Complementation Plan." Although someone makes a mess in the middle, it is the orange juice scene that many people have seen. , , , And its root , which we commonly call Won-na-bul, had a royal road story flow in which the 'chosen one' realizes his calling and power and works tirelessly to achieve a purpose that matches it or to punish evil, and made many people dream and feel excited. As the years passed, if you look at recent popular works such as , , , and , the training scenes are very simple or pass quickly. In some ways, this part can be boring, and in the end, people want to see extremely strong beings fighting or fierce conflicts, so naturally these kinds of stories became interesting. In fact, what I find interesting is that the works that are popular these days fight by drawing the opponent into their own space (philosophy/ego/belief). This is similar to the unique barriers in the Type Moon series that were treated as minor in the past. This is a somewhat different approach from the existing approach of jumping into the opponent's space and blending in or settling in. Confirming the era of division It is no exaggeration to feel that such a strong mental barrier has been cast over the real world as well. The tendency to judge differences immediately without listening to the other person's opinion is found everywhere. YouTube and SNS algorithms that show only similar thoughts to mine with a few clicks have become a kind of 'real-life AT field' , blocking the infiltration of other information or perspectives. It is easy to become trapped in one's own worldview and regard the other person as "a being that does not need to be understood", as if imprisoning an enemy in a territory and pushing forward with overwhelming logic or emotion. Due to this digital environment, the points of contact for conversation are reduced, and as division deepens, we lock ourselves in an isolated 'wall' like the symbol of the anime. Personal life and social risks Living with a mental barrier or a self-imposed boundary may seem comfortable. There is less chance of getting hurt, and you can be confident that your thoughts are always right. However, as the barrier gets higher, the trust that the community has maintained quickly disappears. As the fear that “the other party will attack” increases, even the minimal sense of safety that each person has is broken. The social and political messages delivered in the recent Super Bowl performances and the ideological conflicts that occur in the YouTube comment section in Korea clearly show that this division has already become routine. Dichotomous conflict creates cracks everywhere in the public sphere, shaking the foundations of individuals and society at the same time. The necessity and caution of strong public power At this point, voices naturally arise saying, “Shouldn’t laws and systems step in?” Acts that intentionally cause social chaos through fake news or maliciously distorted information should be severely punished. Just as the angels invaded and threatened the world in Evangelion, unverified conspiracy theories and instigation in reality also erode the foundation of democracy. In order to control this, the state’s public power needs to be strongly exercised. However, if we try to forcibly break the inherent barrier or territorial development, there is a risk that even freedom of expression will be violated. If public power leans toward a specific ideology or loses transparency, it will only cause another conflict and deepen the divide. The misunderstanding here is that strengthening public power does not mean the revival of the White Bone Corps or the riot police. It is about social respect and trust. In the first place, if we call the police “spy” or “inspector” and call the prosecutors “rice-checkers” and dig up the personal information of judges, who would want to sit in that position? If excellent personnel avoid important positions, who would take those positions? Personal Attitude Beyond Fake News No matter what the system, the power to break down an individual's 'mental barrier' is bound to be limited. The real problem is that we already consume only biased information within our own self-imposed boundaries and do not even try to understand others. Therefore, each person needs to have a minimal attitude of 'listening' and 'conversation' before public procedures or legal punishment. Even if others' thoughts feel absurd or unpleasant from my perspective, it is important to have the habit of asking, "Why did they come to this conclusion?" This does not mean that we should unconditionally accept it, but rather that we first try to understand their worldview. Only then can we go through the process of refutation and verification instead of hatred and exclusion, and especially, we can be reborn as a community with self-correcting abilities, not a community where fake news is rampant like it is now. Of course, this process requires a lot of attention. It will take a lot of time. We usually believe and move on from things that are agreed upon in a philosophical manner. For example, when a doctor says something hurts, we just think, “Oh, that’s so,” and when something is reported in the newspaper, we think, “Oh, that happened.” When a professor talks about a certain field, we listen to him as if he is an expert, and when a judge makes a final decision in the third trial, we have to accept it. However, this is only possible in a society with abundant trust capital. When trust capital collapses, we begin to have doubts like these. "Isn't that doctor overtreating me right now?", "Isn't that article biased and fabricated?", "Isn't that professor really an expert?", "I heard that judge is from somewhere. Isn't he just swayed by regionalism?". This kind of approach is the beginning of shaking the rule of law and society as a whole. This is why I always say that we must be wary of fakes and arrogant people. If someone around us does this, we should stop them or keep them at arm's length. What about society then? Looking at today’s society rushing toward division, you might want to take a step back and strengthen your own boundaries. That way, there will be less conflict and it seems more comfortable. However, if you look at it at the national level, a world where all sides are divided and only the boundaries are strengthened will bring about greater anxiety and chaos. Developed countries such as the United States and Japan have a framework in which they negotiate or continue policies beyond the ruling and opposition parties at least on key national issues. Even if they were usually opposing parties, they become one team for the sake of the national interest. In Korea, when the judiciary and the executive branch are sharply opposed due to political interests, and public opinion differs according to ideology or political faction, there have been frequent cases of conflicts that shook the social system. If such a situation is repeated, the entire country will eventually fall apart. What we must not forget Just as AT fields and unique barriers are cool settings, the sight of someone strongly protecting their own beliefs is also attractive. However, I have always said this about Evangelion, Jujutsu Kaisen, and the Fate series. It poses the question, “Can we really reach the answer by simply excluding or destroying others who enter this area?” Division can sometimes create energy for progress, but we must not forget that fragmentation, which is now reaching an extreme, only destroys many things. In comics and animation, the solution to this problem is simple. You tear the opponent's AT field, kill the caster who deployed the unique barrier, or have a stronger area than the one who deployed the area expansion. Then, the opponent dies or is critically injured. It's a cool and fun setting because the good and evil dichotomy is relatively clear and the work is tolerant of death, but these days, it's scary because it seems like this is becoming a reality. Because it's an important moment... For example, whether the impeachment of the president is accepted or rejected, will the opposition camp remain silent? No matter what decision is made, riots or other incidents will recur, or once again, we will have a situation where public opinion is greatly divided and people will pull each other's pants. And now, it is one of the few in-course sections where the whole world is fiercely competing and can run.

Please keep in mind that my experience is limited to software development, finance, and manufacturing, and I have no experience in other fields, so please read with that in mind. 1. Background of the discussion on the Semiconductor Special Act (exception to the 52-hour workweek) “Long-term immersion is necessary for global technological competitiveness” Business/Conservative View : The semiconductor industry must maintain a global lead, so long-term immersion in the research and development (R&D) phase is unavoidable. He says that Korea should also flexibly operate its R&D personnel by lifting the 52-hour limit, at least while citing the fact that “overseas competitors engage in a speed war by conducting all-night research or ‘TF’-style projects when necessary .” “In particular, industries such as semiconductors that require repeated competition in fine processes, optimization of production lines, and design tests often have to focus on this for a week,” is the logic. “Increasing working hours does not lead to creative innovation” Labor circles/some experts : On the other hand, they point out that “creativity and innovation come from work efficiency, quality of immersion, and organizational culture rather than ‘time.’” Even an expert from Samsung Electronics Semiconductor (interview with Director Park Jun-young) emphasized that “long working hours only accelerate burnout and talent drain, and that work distribution and efficiency improvement are important.” Although SK Hynix has applied for special overtime work much less often than Samsung Electronics, there is also a counterargument that “ there is little direct correlation between working hours and semiconductor competitiveness ,” citing recent cases in which SK Hynix has overtaken Samsung Electronics in some technological fields. 2. The old debate of “hard work vs. efficient and creative work” Concerns that “Koreans don’t work as hard as they used to” Among some venture capitalists (VCs) and entrepreneurs, the issue of ‘laziness’ is often raised, with statements such as “These days, startup employees don’t even work 40 hours a week, let alone 52 hours,” and “There are hardly any offices with lights on on weekday evenings. ” “Startups start out on an uneven playing field, so it’s natural to work efficiently and you absolutely have to invest a lot of time .” In other words, the logic is that long hours and high intensity work are inevitable for small organizations to achieve results in a short period of time. “Long hours are not the answer. A culture of efficiency and immersion is the key.” On the other hand, the equation “long working hours = productivity” itself is seen as an outdated paradigm. “There are many research results that show that long working hours actually hinder creative ideas and high concentration, and cause burnout, which also lowers the overall performance of the organization.” There is also a point that “even if it is a startup, a smart work system centered on performance must be established as a basic principle. The probability of success does not increase simply by sitting for a long time.” 3. Various perspectives surrounding “Is the 52-hour workweek relaxation really the answer?” Summary of corporate and government claims In cutting-edge R&D fields such as semiconductors, batteries, and bio, it is important to push ahead at a specific time, so let's apply an exception to the 52-hour limit.

TL;DR The excessive fears and extreme narratives of “loss of hegemony” that exist in the market overlook , above all, the “complex reality” of the AI ecosystem . There is much to learn from the cost-effectiveness and novel approaches of the models being put forth by China , and the United States ' leading position in large-scale model research and computing infrastructure cannot be ignored. Ultimately, as technological and business convergence continues, there is a possibility that the AI competitiveness of both countries will grow together. I would like to emphasize that a balanced understanding based on facts is necessary. The claim that “China’s AI hegemony has surpassed that of the United States” is an exaggeration. China has had AI capabilities from early on , but American Big Tech (OpenAI, Meta, Google, xAI, etc.) also have a globally unrivaled research and development infrastructure and diverse AI ecosystem . It is true that Chinese companies are catching up quickly in autonomous driving, computer vision, and image AI, and startups like DeepSeek are also coming up with innovative models . However, this is not a basis for concluding that “the United States is falling behind” or that “hegemony is completely shifting.”* For example, GPT-4o/o1, Gemini, and LLaMA3.2 still maintain high performance and are being used and tested worldwide. DeepSeek also benefited from Meta's open source policy, including Pytorch and LLaMA. In fact, Meta and xAI have also released models and services with meaningful performance, such as Meta AI and Grok, but I am curious about the criteria for these models that have become such a hot topic. Conclusion : Although China continues to demonstrate strong competitiveness in the AI field, this does not mean that the US-China competition is intensifying, but rather that one side is winning unilaterally. The claim that “Big Tech’s astronomical capex is a waste” is also a simplistic view. The HPC (High-Performance Computing) infrastructure, cloud ecosystem, and large-scale GPU ownership that American big tech companies have built through large-scale capital investment have enabled ultra-large model training and global service expansion. It is true that models such as DeepSeek, Qwen, and TONGYI can achieve sufficient performance at relatively low costs , but this is because they utilize various optimization strategies (MoE, RL, Distillation, etc.). Additional fine-tuning can also be quickly applied, either in combination with open-source models or via techniques such as Low-Rank Adaptation ( LoRA ). In fact, I have been consistently emphasizing Chinese artificial intelligence since 2023, but it has not received much attention. The funny thing is that the most popular model in 3blocks is the LLaMA series, but I pushed Qwen, but it was not chosen much because it is a Chinese model. This is even though we at 3blocks only supply closed models that are installed locally. Large-scale investment itself cannot be seen as a “waste,” and as other approaches gradually emerge and “investment efficiency” increases, each company continues to compete in different ways . Bottom line : The resources and know-how of American Big Tech, built up through billions of dollars of investment, are still valid. The dichotomy of “China did it for less, so American investment is all a waste” is an overinterpretation. The statement that “the scaling law is wrong and other methods are the norm” is close to being misleading.

In the technical world, where wordplay often occurs, completely different meanings are often conveyed depending on how you hear it. This becomes even more evident when complex technical concepts are compressed into short sentences. When the implications or background context of the technology itself are overlooked and only provocative expressions are highlighted, it encourages misunderstanding and even causes unnecessary fear in the reader. If this situation is repeated, the reader will perceive the facts in a distorted form, and the risk of making wrong investments or decisions increases. Mark Zuckerberg's AI remarks are a representative example. He predicted that by 2025, companies will have AI that can perform the role of mid-level engineers. He meant that AI can play a big role in the process of writing code and operating efficiently. He added that this will give humans the freedom to focus on more creative and experimental areas. However, some media outlets changed this into a fear-based message that "AI will replace humans." They reported it as "human engineers will soon lose their jobs" while ignoring the exact context of the remarks, which inevitably leads readers to feel an excessive sense of crisis about AI. The original Joe Rogan podcast was 2 hours and 50 minutes long, and it only took about 2.5 seconds to read the provocative headline... so that's possible, but... Similar distortions and exaggerations occurred in the field of quantum computing. Although there were clearly meaningful research results, such as Qubit and Google's quantum chip Willow, this did not mean immediate commercialization or generalization. Nevertheless, as the prospects for huge results were inflated as if they would appear in the near future, investors flocked to unproven companies. Meanwhile, an abnormal phenomenon occurred in which the stock prices of companies with small market capitalizations or sales fluctuated more than those of large companies that actually possessed the technology. In the end, many people suffered losses, and the correct understanding of technological development was put on the back burner. Some people may have looked at quantum-related stocks as theme stocks, but if you search for quantum computing-related stocks on YouTube or various portals right now, there are many places where they are cleverly mentioned in the form of scientific or informative articles. (Someone who runs a famous YouTube channel published a book, but he has no basic knowledge of physics and does not even understand what quantum computing is.) One thing that should not be overlooked in this process is the responsibility of the information provider. Responsibility does not mean that it should result in legal punishment. Rather, it is closer to the responsibility of acknowledging and accepting one's own comments. In other words, if you misinterpret or relay information with excessive confidence, you must clearly state why you reached that conclusion and why you reported it that way. Here, it is essential to be honest about your own errors in judgment. If you are relaying information created by someone else, you must carefully check whether you have been faithful to the original text and have not distorted the context. If there is a part that has already been relayed incorrectly, it would be your duty to the readers to inform them of the fact and go through the process of correcting it. You should either only say what you know or admit and widely inform if you are wrong, but I have rarely seen cases where this is done. Even people who are not like that become famous or change at some point. Even from the perspective of receiving information, if you spend your money, time, and energy, I hope you look up the original text once in a while. Of course, this is very difficult. If everyone could do this well, there would be fewer people who would play with words like this in the first place... As I always say, most people don't read the text and don't want to look it up. The biggest problem that arises when there are many people who play with words like this is not the increase in fraud victims, but the members of society becoming suspicious of each other. Furthermore, most of the information coming out of the media will not be believed. This will only increase the cost of maintaining social cohesion and social trust. What kind of waste is this, and who is it for? In the end, words and writing have great power in themselves. In complex and specialized fields such as technology, if the context of words is not properly understood, it is easy for serious information distortion to occur. Readers must protect themselves by not being lazy in checking, and information providers must be able to check and take responsibility for their own statements. Through this, we can calm the waves of wordplay and distortion, and face the true meaning and value of technological advancement more clearly.

In the past, product managers (PMs) were mainly responsible for process-oriented tasks such as writing specifications and managing schedules. However, as the market environment rapidly changes and customer demands diversify, PMs are now positioned as executors who directly create products. It is an era where they must prove the value of products through the process of producing actual results and receiving feedback quickly, rather than simply coordinating plans on paper. Prototype: An Essential Tool for Validating Value At the heart of this change is prototyping. In the past, it was common to organize ideas and list estimates in documents. However, users feel much more intuitive value from a working prototype than from a plan explained only in text or tables. That’s why many PMs use no-code and low-code platforms to quickly create an MVP (Minimum Viable Product). Through this prototype, they can check the market response in real time and find and improve problems that were missed at the beginning. This process also greatly helps to increase the completeness of the product and bring forward the release time. Differences in PM roles in large companies and startups In large companies, PMs are often tied up with paperwork and presentation-based work because decision-making processes are complex and reporting lines are long. A culture is established where they must focus on managing schedules and quality within a limited scope while coordinating the interests of multiple departments. On the other hand, startups emphasize the ability to quickly implement ideas, release the results to the market immediately, and receive feedback. A single prototype can be a direct path to proving the potential of a product, and in many cases, it can lead to new investment or partnership opportunities. This does not mean that large corporate culture is necessarily behind. Even within large corporations, they are operating Agile teams or internal venture projects and trying to introduce speed and flexibility similar to startups. Through this, large organizations are trying to obtain both agile execution and innovative ideas at the same time. Technological advancements and the expanded role of PM As AI and automation tools are actively utilized, the technical understanding that PMs must have is becoming increasingly important. In particular, as no-code and low-code tools become popular, an environment has been created where even non-professional developers can create initial prototypes themselves. This greatly reduces the time and cost required to verify ideas, and AI-based analysis tools can be used to quickly and accurately identify user behavior data. As a result, it is also advantageous in determining the priorities of product improvement. Of course, not all PMs need to write code. However, if they can handle simple modifications or UI prototyping themselves, communication with the development or design team becomes much smoother. This ability to reduce collaboration costs is highly valued regardless of the size of the organization. PM now covers more As I mentioned in my previous post, PMs will no longer be limited to planning and management as they were in the past. They will also create prototypes and immediately reflect feedback from the market and customers to verify value. This has become a key competitive advantage not only for startups but also for large companies. The ability to create products that work in the market faster than anyone else and to immediately reflect feedback to improve the product is the most sought-after quality for PMs these days. And as the technical capabilities that support this become increasingly important, the role of PMs is also evolving in a broader and more challenging direction.

There is a proverb that says, “A word can pay off a thousand nyangs.” On the contrary, a word can destroy a relationship. In our daily lives, we exchange thoughts and feelings through conversations with others, but we also experience regrets like, “Why did I say that?” As Christmas Eve approaches, I will take out one of the writings that I have been putting off. This is a mistake I also make a lot when I am having a conversation. As I always say, if the listener does not understand, it is largely the speaker’s fault, so I would like to talk about things that ruin conversations centered on the speaker. In fact, there are common patterns in the language that people use in conflict situations . Even if you think you are using logical and objective language, when you look back, you often find that you are using aggressive language because of your agitated emotions. That is why psychologists call this “ destructive communication patterns .” Six typical destructive patterns (judging, blaming, forcing, comparing, taking for granted, and guilt (rationalization)) can easily be incorporated into our words, cutting off our “connection” with others. In this article, we will look at how these six patterns manifest, why they are problematic, and how to overcome them. Of course, if each of us avoided these patterns, wouldn’t we all fight less? Judging Definition : A speech pattern that pre-judges and presumes the other person's behavior, personality, motives, etc. Example sentences “That’s just how you are.” “I knew it, Bona Mana.” “You won’t change anyway.” Judgment may seem like a simple evaluation on the surface, but in fact, it does not open up the possibility of change and communication to the other person . With just one phrase, “You are just like this,” the conversation partner feels helpless, feeling, “I am already labeled.” This is the most commonly used pattern anywhere in the world. It is a way of thinking by prejudging based on race, country, language, culture, gender, age, etc. Ironically, everyone feels uncomfortable being categorized and judged in advance like this, but in fact, there are many cases where they themselves do this. In psychology, this kind of 'judgmental' language reinforces fixed mindsets . According to Carol Dweck 's research, the attitude of defining others as "fixed traits" hinders cooperative relationships and often causes conflict. When one side insists, "You can never change," the other side either gives up trying to change themselves or only increases resentment. How to improve Speaking in a situational sense Instead of saying, “Why are you always late?” point out a specific situation , such as, “I regret being late today and missing an important discussion.” Be curious If you approach it with the attitude of asking yourself, “Why did you do that?” you will explore the cause together rather than making conclusions. Blaming Definition : A language pattern that directly attacks the other person or unilaterally shifts responsibility. Example sentences “It’s all your fault that I’m in this state!” “How come you cause trouble every time? You really have no answer.”

The speed of innovation achieved by the meeting of technology and AI is truly dazzling. Artificial intelligence, which has naturally taken root next to us, is now changing our daily lives in various ways, beyond text, through images, videos, and voices. In particular, the series of updates from OpenAI, which were announced sequentially from December 5 to 16, are attracting much attention because they show this trend in a condensed form. Following the launch of ChatGPT Pro, various contents such as reinforcement learning-based fine-tuning research programs, text-based video production platforms, developer tool expansions, and next-generation model demonstrations are pouring out every day, suggesting that a “comprehensive AI ecosystem” is being built. Before introducing the full functions, let’s briefly look at the meaning of this massive announcement. Day 1: o1 and ChatGPT Pro Launch ChatGPT Pro Launch ($200/month) O1 and o1-mini model access rights Real-time web search Advanced Data Analysis DALL-E 3 Integration 32K token support Day 2: Reinforcement Learning Fine-tuning Research Program Announcement of a research program for fine-tuning reinforcement learning Alpha access for researchers, universities, and corporations Support for customized model development Real-time performance monitoring Providing API integration environment Day 3: Sora Release Sora Video Creation Platform Unveiled Create high-quality videos up to 1 minute long Text-based video production Edit/enhance existing videos Building on DALL·E technology Day 4: Canvas feature updates

Artificial intelligence (AI) is shaking up the world. The vast amount of data that was once simply accumulated has now become a key driving force that is reorganizing the entire economy and society. In the past, information was only meaningful when it was “stored,” but now “learning and utilization” determines its value. In this transition, traditional copyright-based profit models or data sales strategies that depend on the number of API calls are gradually revealing their limitations. The concept that is emerging at this point is “data licensing.” This defines data as an evolved concept that encompasses utilization, conditions, and new value creation methods, rather than a simple purchase and sale target. Through data licensing, companies are moving beyond simply distributing content or providing APIs, and are attempting to structure data into a form suitable for AI learning, thereby building a more flexible revenue model. In particular, we are now in an era where multi-modal data such as images, voices, and videos are being fully utilized. It is essential to move beyond simple text tagging and attach meaningful and usable labels to various media so that AI can understand and apply them more precisely. This will allow AI models to grasp richer contexts and conduct in-depth learning with limited data. This change will provide new business opportunities for data owners, while also providing unprecedented diversity and scalability to data users. Limitations of the existing model: Copyright and API-based revenue structure issues Traditionally, data-based monetization strategies have been largely divided into two axes. First, there was a method of generating revenue by selling or streaming existing content such as music, videos, and documents based on solid copyright. Second, a model was widely used that provided APIs that could access specific data and charged based on the frequency of calls or usage . However, these two models exposed the following limitations in the AI era. Limitations of scope of application : The copyright-based model provides clear rights protection, but pre-determines the scope of data use. This limits data reprocessing, reuse, and creation of new forms of content. On the other hand, the API-based approach charges per call volume, so extensive data use can easily result in a high cost structure. Complex trading processes and high barriers to entry : The complex intertwining of contracts, licenses, and authorization processes for data use makes it difficult for companies and institutions to easily access it. This especially limits access to high-quality data for small startups or research institutes. In other words, it acts as an initial barrier to entry for those who want to develop high-value AI models. The gap between data value and use : Data owners own the content, but the real value lies in the new insights, predictions, and service improvements that AI models create by learning from that data. It is difficult to maximize this “utilization value” by simply selling ownership or granting usage rights. The Meaning and Background of Data Licensing Data licensing is gaining attention as an alternative to overcome these limitations. It is a new monetization strategy that goes beyond simply “the right to provide data” and allows data users to reinterpret and process data according to various conditions and purposes . In simple terms, data licensing can be said to be a “customized” supply contract that opens up the possibility of data utilization. Recently, Reddit and American media companies have created a new revenue model by providing crawling to companies conducting artificial intelligence learning more easily through data licensing contracts. Of course, community platforms like Reddit are also at the center of this, which has provided a completely new opportunity to those who were only considering the existing traffic-based advertising revenue model. (Closing a deal worth hundreds of billions of won) Data licensing is a different concept from copyright, and the differences and characteristics are as follows. Division Copyright Data Licensing Definition Copyright is the legal right that a creator has over his or her creative work, allowing him or her to control the reproduction, distribution, and performance of the work. It protects the form of expression of the creative work itself. A contract that grants permission to use data or information, allowing the use of a database or large data set. It protects the collection or structure of data, and is particularly relevant to the use of data for training AI models. Object of protection A form of expression of creative works such as literature, art, music, and video. A collection of information, such as a database or data set. When rights arise

I have been doing personal reflections consistently since 2015, and from 2016 to 2023, I have been running a public reflection program called 'RetrospectHaebom' and sharing moments of growth with many people. However, I have decided to no longer conduct this public reflection activity. However, this does not negate the value or necessity of reflection itself. First of all, I want to make it clear that the act of reflecting is a very meaningful and wonderful activity. Through reflecting, we can calmly organize what we have done in the past, and clearly recognize what was good and what needs to be improved, what we have achieved and what we have given up along the way. Based on this, we can design the future or reinterpret past experiences. Some people may express regret that they do a reflective activity once a year, or that they try to do it briefly and then quickly give up. However, regardless of the frequency, whether it is a short or long time, or even for the purpose of showing it to others, reflecting is meaningful in itself. The important thing is the process of being honest with yourself, and through this, you create an opportunity for reflection. It is also attractive that there are various retrospective techniques and methodologies. From KPT (Keep-Problem-Try) to time series techniques and photo association methods, you can enjoy retrospectives according to your own tendencies and situations by utilizing various approaches. I have also been conducting retrospectives with many people for over 5 years, discussing better retrospective methods, and sharing new perspectives and experiences. In this process, getting a glimpse into other people’s lives and gaining insight has been a great inspiration to me. However, what I am now going to stop is the 'public retrospective with many'. This does not mean that I am giving up personal retrospectives, but that I am stopping the retrospective activities that were conducted in the form of public programs. The reasons are as follows. The rise of paid retrospective programs : Paid retrospective services targeting young professionals and juniors are on the rise. This is creating a tendency to make retrospectives into a kind of ‘product’ rather than a simple ‘place for reflection.’ Sales practices under the guise of retrospectives : There are increasing cases of selling lectures or templates under the name of retrospectives, which is creating a commercial movement that blurs the inherent value of retrospectives. Excessive authority : There are also instances where the act of reflection is overly glorified or given excessive authority, such as “you have to reflect to be mature.” This can cause reflection to be seen as a “must-do” or “evaluation” for some people, rather than a free and honest form of self-reflection. In the first place, looking back is something you can do with just a fan and a notebook or a laptop or a smartphone. I always say that instead of spending money on strange things, there are many ways to use it for more valuable things, such as having a year-end party with friends, buying meat, etc. Of course, if you don't spend money, you can do it... but looking back on the past few years, if you don't share the year sincerely and do it sincerely, this act of looking back can end up as a formality that just makes you feel good. The reason I am stopping public retrospectives is because I no longer want to watch the voluntary meaning of retrospectives being damaged in this flow. Retrospectives, which should be honest and free reflections for individuals, are commercialized or authoritative at some point, which goes against the values I wanted to share through retrospectives. Also, the moment retrospectives themselves are seen as something that costs 'money', I feel bad about it. Is it something like Hongdae disease? It seems like I have a personal aversion to selling things that are easy/obvious, just like with Notion . That doesn't mean I'll stop doing retrospectives. I'll continue to do them on a personal level, and through them I'll continue the process of looking back on my life and finding clues for growth. However, I'll stop doing them publicly like I used to, or leading others' retrospectives and interacting with them. I'm going to use retrospectives in a more personal space and way. By doing this, I hope to once again regain the free reflection and honest conversation that the act of 'retrospection' originally contained.