{"slug": "how-humans-reason-different-from-ai-and-why-thats-an-investment-opportunity-yoon", "title": "How humans reason different from AI and why that’s an investment opportunity | Songyee Yoon interview", "summary": "Former NCSoft president Songyee Yoon, now founder of Principal Venture Partners, argues that AI learns in a broken way compared to humans and that this gap presents an investment opportunity for startups. Yoon, who holds a Ph.D. in Computational Neuroscience from MIT, says Silicon Valley's assumption that removing friction is inherently good is flawed, and she advocates for treating AI as an algorithm rather than anthropomorphizing it. Her fund focuses on AI-native companies, and she has invested in Together.ai, Cartesia, and Sesame.", "body_md": "*Become a member of GB MAX to gain exclusive access to the industry and to the most influential global B2B leadership community in the business of gaming, entertainment, and tech.*\n\n[Join now](https://go.gamesbeat.com/gb-max/)and also get a VIP ticket to GamesBeat Next (Nov 2-3, SF).[Songyee Yoon](https://www.songyeeyoon.org/) is the former president of [NCSoft](https://us.ncsoft.com/en-us), the South Korean maker of online games including Aion, Guild Wars, Guild Wars 2, Blade & Soul, Lineage, Lineage 2, Tabula Rasa, City of Heroes and Fuser.\n\nThree years ago, she left that role and moved to Silicon Valley to focus on a passion for technology, namely AI. Yoon, who has a doctorate from MIT and earned a law degree during the pandemic, started a fund to invest in AI-native startups.\n\nYoon is the founder and managing partner of [Principal Venture Partners](https://www.principalvc.com), a VC fund focused on investing in AI-native companies. She’s also a board director at HP and a member of the MIT Corporation (MIT’s board of trustees). Prior to founding PVP, she spent nearly two decades immersed in AI and games at what became one of the world’s largest gaming companies, NCSoft. She’s also an early investor in companies like Together.ai, Cartesia, and Sesame.\n\nYoon often got asked by parents who wondered if games are good for children. So much so that she wrote *Push Play: Gaming for a Better World*, which documents how play is in our DNA.\n\nAlways focused on science, technology, engineering and math (STEM), Yoon graduated from the Korea Advanced Institute of Science and Technology and went on to earn her Ph.D. from MIT in Computational Neuroscience.\n\nYoon recently published an essay that shares the perspective she brings that we don’t see often in this space. “[Friction As a Feature, Not a Bug](https://www.principalvc.com/blog-post-entry?slug=friction-as-a-feature-not-a-bug)” argues that Silicon Valley has been operating on a flawed assumption for decades, that removing friction is inherently good. Yoon believes that AI currently learns things in a broken way, as revealed in our conversation. Humans are much more natural learners, as we can reason more quickly and intuitively. That’s why she thinks there’s a lot of opportunity to fix AI for the better through startups.\n\nWe talked about the distrust of AI, and possible differences for why Asian countries are welcoming AI now and why Western countries are resisting it and treating it with skepticism, as well as why the center of AI is in Silicon Valley.\n\nOne of the problems is that we anthropomorphize AI she said.\n\n“We talk about AI safety, as if AI were a threat. In fact, it’s just an algorithm,” Yoon said.\n\nIt is used by humans. We should talk more about human safety, or how we can govern the humans who use AI, she said.\n\nHere’s an edited transcript of our interview.\n\n**GamesBeat: How have you navigated the different moves you’ve made over time? When did you move to your newest position?**\n\n**Songyee Yoon:** It’s been about two and a half years. I left NCSoft about three years ago and moved on to starting up this fund. We started investing about two years ago. We opened the new office earlier this year in Palo Alto. We’ve been deploying capital for a couple of years.\n\n**GamesBeat: What motivated you to make the change?**\n\n**Yoon: **My background is AI. That was my PhD. I’ve been using it. I was very interested in that part of the development process at NCSoft, and I worked with it in other industries as well. I felt that this was a great time to start new companies. Throughout your career, I’m sure you’ve seen that there’s been a lot of promises, a lot of disappointment, a lot of underdelivery and so on. But we now finally have capital, talent, and data coming together to really build something useful. It’s very transformative. Just like we saw in 2000, when everything went digital, internet, online. It’s comparable to that. I felt like it was a great time to be part of that transformation.\n\n**GamesBeat: What are you zooming in on within AI?**\n\n**Yoon:** Despite the massive progress we’ve seen, it’s still very early. There are a lot of things to get excited about. It’s in the nature of the technology that I’ve seen. People tend to be very excited about it. Maybe because of the name, maybe because of the anthropomorphization of the technology. But somehow I think it’s more–it’s undoubtable that there’s a transformation that will come, but it will be a bumpy road. Some of the things we see today will become obsolete. Some of them may not become part of the standards that eventually roll out.\n\nAs a fund manager today, I’m trying to pick companies that can endure that bumpy road and continue to be relevant, regardless of what happens at certain layers of the technology’s development. That’s what I do. I try to find infrastructure layer developments that are going to be relevant in the coming decades and make investments in those companies.\n\n**GamesBeat: I feel a little similar in that I’ve watched a lot of the AI investments in games, and most of them are doing either generative AI art or other kinds of tools to make game development more efficient. But I’ve seen a couple of companies start making money from their data. Arkadium was a little puzzle game company that’s been around for 25 years. They kept all of their gameplay data and used that to learn what kind of games to make. It turns out that data is pretty useful for training models to reason and play games like Tetris. They did some research, and they found that by the third move of a game like Tetris, the AI would make a bad move that no human would ever make. A multi-billion-dollar model would drop a piece upside down or something like that. They’re licensing their game data out to the model companies now, and so the models are getting better.**\n\n**It’s the same with Medal and General Intuition. They just raised a pretty big round, $320 million, to make use of the gameplay data they record as people upload their favorite moments in games, victory clips and things like that. This turns out to be useful for real-world robotics. It’s funny how these turned into revenue-generating companies while we figure out more things to do with AI. That surprised me. Do you find that kind of thing is also what you might be looking for?**\n\n**Yoon: **It’s not quite what I’m looking for, but it’s very insightful. It tells us a lot about where the technology is. There’s a blurry line between research and commercialization. We’re very rushed in terms of bringing new breakthroughs and innovation into the commercial space, trying to build companies around those ideas. One of them is this notion of zero to one learning. What you said there about puzzle games is absolutely right. When people play Tetris for the first time, you don’t have that kind of process where you do something completely silly, run that back many times, and eventually you get better. That’s not how you do it. You see it, you learn from it, and some people can get it right and do very well after only a couple of tries. It’s a different mode of learning that you see in humans than how general AI architecture is built and how it learns at the moment.\n\nThe implication of that is–there’s still a lot of research going on in academia about what an alternative architecture could be. What architecture would allow this kind of one-shot learning? You don’t have that long process to get there. You see it and you get it. That’s one of the reasons we’re very early in terms of AI. The architecture that is most suited for a certain type of task and problem is yet to come. That’s one very good example, a door into that insight.\n\n**GamesBeat: Pulling back more, how do you get past some of the resistance a lot of people have to AI right now? There’s a general sentiment about how AI might take away jobs, but the more specific thing we see in the gaming audience is, “Well, this is producing nothing but AI slop.” These one-shot games may sound promising, but they turn out to be problematic in the way they maybe copy something else too much, or they’re just not holding anyone’s attention. Game developers, meanwhile, are saying that they don’t have tools yet that can really help them.**\n\n**The resistance we see is interesting, because the game industry historically has always embraced technology. It adopted CD-ROMs first, for example. But now we’re in this moment where it feels like the game industry is going to lag. Perhaps this is only happening in the west, but I wonder how you think people can get past this and get to embracing the technology.**\n\n**Yoon: **When it comes to AI slop, there’s no audience for it. That business isn’t sustainable. The technology may not be there to be comparable to a human yet, but I think the market will tell. Games will find traction, find players, and studios will make the choice to make investments where it matters, in what players choose to play. Gaming is a business. The market mechanisms are there.\n\nThe more important thing here is that resistance to AI technology–it’s somewhat misleading. The term “AI” itself is misleading, because it leads people to anthropomorphize the technology very easily. We sometimes use terms like “AI safety,” as if AI were something inherently dangerous. But it’s just technology. It’s an algorithm. It’s the humans who use it. What is the right governance mechanism for them? What are the right checks and balances to have in place? What topics are policy makers paying attention to, like data integrity and training and optimization and accessibility? There are many issues that have nothing to do with the technology itself. It’s about the humans who are using it. Who owns it? Even though the term has easily swayed us to think otherwise, we should keep our eyes on the fact that it’s really humans who are using this in a suboptimal way.\n\n**GamesBeat: Where do you see the most progress being made with the use of AI? Are there already big efficiency payoffs in areas like coding?**\n\n**Yoon:** When people think about AI, they have a certain image in their minds. “This is AI.” A lot of people think a chat bot is AI. But AI is much more than that. It has many different forms with many different ways to use them, many different ways to integrate them into the enterprise. There’s a lot of attention on using more data, more software, more automation of repetitive work. That will come. Whether it’s using big tech solutions or custom solutions, that will become easier and more affordable. That’s the trend. How it’s going to be implemented over time and what kind of guardrails will be put in place, those are our choices to make.\n\n**GamesBeat: Where are the openings for startups? I see startups being formed still that are saying they’re going to knock Nvidia off the top. It’s hard to see an opening in that kind of area. Maybe they hope to be bought by Nvidia someday. But I wonder what might be an easier business to find a space in.**\n\n**Yoon:** It’s already happening. Nvidia is a tremendous company with a supply chain that’s very hard to replicate. They have the CUDA software platform and so on. But there are many startups, whether it’s a software play, whether it’s semiconductor design, whether it’s memory chips for certain kinds of computation–there are many kinds of innovation coming out of the startup ecosystem. We’re very excited about those.\n\n**GamesBeat: How does your gaming background help you in this role?**\n\n**Yoon: **Nvidia started by providing gaming chips. It’s not a coincidence. Gaming has always been an area that’s at the forefront of innovation. We hear a lot about AI adoption because gaming as an industry tends to be very innovative and embrace new technology. That’s also because our audience is early adopters. Gamers themselves are early adopters. They’re interested in new technology and what’s possible, what’s out there. If you pay attention to gaming you naturally develop a handle on the early adopter audience. That culture is baked into gaming. It helps us to see what kind of technology is relevant, what has commercial potential.\n\nGaming is an international business as well, on a massive scale. Its customers are very particular about things like downtime and latency. All of those things matter to them. We’re used to paying attention to those particulars. With the adoption of new technology, more people will care about the things that gamers have been caring about for a very long time.\n\n**GamesBeat: Was there a reason you chose Silicon Valley as the place to be?**\n\n**Yoon:** I love the community. It’s full of builders. At one point, just in the city of San Francisco alone for one weekend, there were more than 250 hackathons happening simultaneously, small and large. What other city is out there with that many interested builders who are coming together and doing things together, hustling and hacking? That’s the real engine of innovation. I wanted to be surrounded by that community.\n\n**GamesBeat: You have experience in Asia as well. I remember talking to Yat Siu from Animoca Brands. He always had an interesting perspective. He felt that with blockchain technology, Asian countries embraced it earlier and faster. They were trying to figure out how to bring it into games. Asian audiences gave it a chance. They weren’t as naturally resistant to it as western audiences were, just as is happening now with AI. Having a foot in both of these worlds, do you think Asian technologists and Asian audiences are embracing technology faster these days, while the west puts up this wall of skepticism?**\n\n**Yoon:** It’s not black and white. I don’t think you can just say that people are more innovative in Asia versus the west. It’s more subtle than that, more nuanced. What I can say is–if you go to South Korea, there are still a lot of people there who experienced the war. It was the 1950s. Everything there has been built since then. People have come through that massive transformation in their lifetime. They can see, they know from their experience, that that level of transformation can happen.\n\nIn more peaceful parts of the world, the ways of living, the daily routines haven’t changed as drastically. There’s not as much incentive to take the leap, to go to the other side. It’s a tremendous risk. What you have right now is already very good for you. Why would you give that up and take a big risk where you might lose everything? I think there’s a cultural fabric that’s been built up because of the very compressed timeline of history that people have lived through in east Asia, as well as other cultural differences that have contributed to that appetite for taking risks.\n\n**GamesBeat: San Francisco and Silicon Valley took a big blow in the pandemic. All of a sudden everyone started working remotely. They figured out that they didn’t have to be in downtown San Francisco to be in the right place. Remote work has spread far enough that I also think–maybe it doesn’t make sense for any one place to be dominant when it comes to creating new technology. But at the same time, housing prices in San Francisco are still going crazy. Everyone who’s gotten wealthy in AI is moving there. It’s coming back stronger than ever. There are these contradictions in the way work can be done anywhere now, but people still gravitate to where they think the center is.**\n\n**Yoon: **It’s a good point. There’s a value in getting together in person. There’s a lot of knowledge that people can pass along that they might not be ready to articulate in writing. I get inspired by interacting with other people in person, getting their unpolished thoughts and picking up on where they’re coming from. A lot of people here in Silicon Valley value learning from each other a lot.\n\n**GamesBeat: Are you announcing any of the details about the fund so far – how much money you’re investing, how much the average investment might be?**\n\n**Yoon:** We’ve already started investing in companies. We have more than 20 portfolio companies. We’ve invested from seed to series B. We’re very excited about everything our founders are doing.\n\n**GamesBeat: Is there any gaming specifically in there, or is it all AI?**\n\n**Yoon:** Because of my background I get to see a lot of innovation in that space. I’m not looking at gaming in terms of creating IP or content. But I’m very interested in innovation that comes out of the gaming ecosystem. The technology innovation we see in gaming, how can that be translated into other businesses? There are a lot of companies working on that, things that are possible and affordable because of AI and the integration of that technology.\n\nAlso, what is the future of gaming? The games we remember have levels and characters and missions and tests. But what’s next? Gaming isn’t just characters and graphics and models. Gaming is engagement and retention. What’s going to be the next experience on the frontier out there that will spark the imagination and curiosity of the audience? That’s a really interesting quest. I’m always on the lookout for what’s next.", "url": "https://wpnews.pro/news/how-humans-reason-different-from-ai-and-why-thats-an-investment-opportunity-yoon", "canonical_source": "https://gamesbeat.com/how-humans-reason-differently-than-ai-and-why-thats-an-investment-opportunity-songyee-yoon-interview/", "published_at": "2026-07-24 13:30:00+00:00", "updated_at": "2026-07-24 16:13:06.861048+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-startups", "ai-research", "ai-safety", "ai-ethics"], "entities": ["Songyee Yoon", "NCSoft", "Principal Venture Partners", "MIT", "Together.ai", "Cartesia", "Sesame", "HP"], "alternates": {"html": "https://wpnews.pro/news/how-humans-reason-different-from-ai-and-why-thats-an-investment-opportunity-yoon", "markdown": "https://wpnews.pro/news/how-humans-reason-different-from-ai-and-why-thats-an-investment-opportunity-yoon.md", "text": "https://wpnews.pro/news/how-humans-reason-different-from-ai-and-why-thats-an-investment-opportunity-yoon.txt", "jsonld": "https://wpnews.pro/news/how-humans-reason-different-from-ai-and-why-thats-an-investment-opportunity-yoon.jsonld"}}