Table of Contents
Artificial Intelligence Translates the Language of Animals
Kim Kyung-jin, Attorney at Law
The Story of AI Learning to Listen to Whales, Dolphins, Birds, and Bees
Twelve chapters on how AI listens to dolphins, sperm whales, humpback whales, birds, and bees to find rules in their sounds, and what this technology means for its risks and for animal rights. Written in simple sentences a child can read, with verified sources in every section.
Table of Contents
AI Deciphers Ancient Scripts
Kim Kyung-jin, Attorney at Law
Ancient Records Revived by AI
In twelve chapters, this book explains how AI revives records once unreadable, from burned scrolls and wooden slips buried in mud to broken clay tablets. It covers virtual unrolling at Herculaneum, virtual collation of oracle-bone texts, reading Silla wooden tablets, computational analysis of undeciphered scripts, and multispectral archives, with verified references for each chapter.
Table of Contents
Artificial Intelligence for New Materials Design and Rocket Propulsion Engineering
Kim Kyung-jin, Attorney at Law
AI Potentials, Self-Driving Laboratories, and Physics-Informed Machine Learning (PIML)
Ten chapters on how artificial intelligence is changing new materials and rocket propulsion: atomic simulation, generative models, self-driving labs, high-temperature alloys, metal 3D printing, combustion, cooling design, and engine diagnosis and control. Written without equations, with verified sources in every chapter.
AI Library
The Age of Autonomous Scientific Discovery
Kim Kyung-jin, Attorney at Law
AI Scientists and Self-Driving Labs
This book follows how AI scientists and self-driving labs are changing the way science generates and verifies claims. It covers literature-based discovery, natural-language protocols translated into robot commands, multi-agent research systems, closed-loop laboratories, materials search, the verification gap, chains of evidence, research harnesses, journal ethics, and legal responsibility.
AI Library
A New Era of Life Sciences Opened by Artificial Intelligence
Structural Proteomics, Genomic Foundation Models, Autonomous Laboratories, and Global Governance
Kim Kyung-jin, Attorney at Law
This book is a research volume compiled with artificial intelligence. A human selected the materials and structured the work, while AI models drafted the sentences and cross-checked the facts.
AI Library
The Double Structure of Digital Sovereignty
Europe’s Departure from Palantir and the Chains of American Big Tech
Kim Kyung-jin, Attorney at Law
This is a record of 2026, when European intelligence agencies and defense ministries began removing analytics tools from America’s Palantir. It covers the replacement decisions made by France’s General Directorate for Internal Security (DGSI), Germany’s Federal Office for the Protection of the Constitution (BfV), and the Netherlands Ministry of Defense; the incident in which US export controls severed an ally’s ac…
New English Edition
Artificial Intelligence in Horticulture
Kim Kyung-jin, Attorney at Law
Across five chapters and ten sections, this book examines computer vision for crop diagnosis, harvesting robots and autonomous field systems, smart greenhouses and digital twins, precision irrigation and supply-chain quality control, high-throughput phenotyping, and predictive breeding.
New English Edition
Artificial Intelligence in Food Crop Agriculture
Kim Kyung-jin, Attorney at Law
Across six chapters and eighteen sections, the book examines digital agricultural infrastructure, remote sensing, crop diagnosis, yield forecasting, precision irrigation, genomics, molecular breeding, agricultural robotics, climate-smart agriculture, and global food security.
New English Edition
The Future of Forestry and Agroforestry
Kim Kyung-jin, Attorney at Law
Driven by Artificial Intelligence and Digital Innovation
Across five chapters and fifteen sections, the book follows satellites, drones, LiDAR, digital twins, forest-specific language models, wildfire and pest forecasting, forestry robotics, agroforestry, timber traceability, and forest carbon markets.
New English Edition
Smart Livestock Farming: AI Enters the Barn
Kim Kyung-jin, Attorney at Law
Sensors listen, cameras watch, and artificial intelligence helps farmers decide.
Across five chapters and fifteen sections, the book follows precision livestock farming from animal health and reproduction to robotic milking, virtual fencing, digital twins, methane reduction, welfare, and data ownership.
Table of Contents
Han Dong-hoon, Busan Buk-gu Gap: A Record of the 100 Days Before and After the Election (Mar. 26-Jul. 3, 2026)
Kim Kyung-jin
Table of Contents and 13 sections
From March 26 to July 3, 2026, this record follows the spring after expulsion, the Busan Buk-gu Gap by-election, victory as an independent, and the first bill submitted in the National Assembly.
Table of Contents
Artificial Intelligence and Medicine
Kim Kyung-jin, Attorney at Law
AI in clinical care, hospitals, education, and research
AI in medical imaging, risk prediction, treatment planning, hospital operations, education, and research, with patient safety, privacy, and accountability.
[AI Library] 23 Gemini and Next-Generation Models
Demis Hassabis, Father of Google's Artificial Intelligence
Part 8. CEO of Google DeepMind
23 Gemini and Next-Generation Models
Kim Kyung-ran, Kim Kyung-jin
At Google DeepMind's headquarters, the lights never went out. Monitors that once displayed Go boards and protein structures were now filled with the parameters of massive language models. Demis Hassabis stood gazing out the window at London's nightscape, lost in deep thought.
He had devoted his entire life to solving scientific challenges, but what he now faced was not a question of science. It was a question of survival. The 'Code Red' triggered by ChatGPT's arrival had shaken the entire ship that was Google, and even DeepMind, once devoted to pure research, had been pushed to the front lines of product development. Hassabis had to turn this crisis into an opportunity.
It was not about building a chatbot. He needed to lay stepping stones toward the artificial general intelligence (AGI) he had always dreamed of. The result was 'Gemini.' Gemini's birth began with tearing down the political and technical walls inside Google.
The merger of Google Brain and DeepMind, longtime rivals, was virtually unprecedented. When Jeff Dean and Demis Hassabis joined forces, it went beyond combining headcount. It was an experiment in fusing two distinct research cultures. Hassabis later reflected: "We combined each side's strengths and gave birth to an entirely new species that had never existed before."
The project name 'Gemini' symbolized the union of two organizations, just as the constellation represents twins. It also carried the spirit of bold ambition from NASA's Gemini program, which paved the way to the Moon. The release of Gemini 1.0 was a signal flare announcing that Google had stepped into the AI competition ring. But what Hassabis truly wanted to showcase was not text generation alone. What he emphasized was how humans perceive the world: processing text, images, audio,
and video simultaneously. He called this 'Native Multimodal' capability. If earlier models were like students taught only text and then forced to learn pictures, Gemini was like a child born with eyes, ears, and a mouth all at once. This difference was decisive.
Gemini showed a remarkable ability to watch videos laden with subtle nuance and read the emotions within them, or to interpret graphs from complex physics papers and derive new equations. The technical evolution moved at a breathless pace. The arrival of Gemini 1.5 Pro brought a revolution in 'Long Context.'
Being able to process one million tokens, and eventually two million, was like giving an AI the memory of an elephant. It meant the system could read hundreds of dense academic textbooks in an instant and locate specific information, or watch a one-hour movie and produce a flawless summary. Explaining this technology, Hassabis said: "AI can now survey the entire landscape of a vast body of information, rather than looking at fragments."
This fundamentally changed how lawyers review tens of thousands of pages of litigation records or how scientists analyze decades of research data. Gemini 3.0 achieved a dramatic leap in the 'Reasoning' and 'Planning' capabilities Hassabis had long craved. It moved beyond a 'stochastic parrot' that merely predicted the next word by probability, becoming a system that thinks on its own and arrives at answers through logical steps. Gemini 3 ran its own verification processes to solve mathematical challenges and, when errors appeared in code it was writing, debugged and corrected them on its own.
This was the synergy created when the reinforcement learning techniques DeepMind had used to build AlphaGo were combined with large language models. Watching it, Hassabis quietly remarked: "We have only now stepped through the entrance to the thinking machine." These technical achievements soon translated into market response.
Monthly active users (MAU) of the Gemini app and website surpassed 650 million, a figure that carries meaning well beyond statistics. It shows that a significant portion of the world's population is asking Gemini questions instead of typing into the Google search bar. Students open Gemini for assignments, programmers for coding, writers for inspiration.
Within Google's ecosystem, Gemini became a personal assistant organically linked to Gmail, Google Docs, and Drive, analyzing user data and automating tasks. This showed that Google had mounted a powerful counterattack against the market ChatGPT had claimed first, armed with its infrastructure and ecosystem. The road was not always smooth. The competition with OpenAI demanded constant tension from Hassabis.
When Sam Altman unveiled GPT-4o and moved on the voice assistant market, Google DeepMind was under pressure to deliver a real-time conversational model that was even more natural and had zero latency. Criticism poured in at times: "Google is too slow," or "Google is resting on past glory." A controversy over historical errors in the image generation feature left a painful lesson for Hassabis and Google.
The incident forced a deep philosophical reckoning, not just about technical perfection but about how AI should understand social context and ethics. Rather than making excuses, Hassabis humbly acknowledged: "We are learning, and we will keep making corrections to build more responsible AI." The Gemini series was the process by which the mission to 'understand intelligence' left the abstract confines of a research lab and became a reality unfolding in the palms of eight billion people worldwide.
He wakes before dawn and checks Gemini usage logs coming in from around the world, witnessing the intelligence he created helping someone code, co-writing someone's poem, and easing someone's loneliness. It is the realization of the AI he had dreamed of on a chessboard as a child: a tool that extends human intellect. Gemini is the largest and most complex tool ever built for making humanity smarter, and the one Hassabis spent 40 years preparing.
Gemma: The Open-Source Lightweight Model. In a corner of Demis Hassabis's office sits a photograph of an old Apple II computer. It is the machine on which he first taught himself to program and became a creator in the digital world. Hassabis often recalled those days and thought that if the technology back then had been monopolized by a handful of corporations and research labs, a boy like him could never have built DeepMind.
This experience played a decisive role in the fierce 'open-source debate' inside Google. "We must share our cutting-edge technology with the world.
That is how a second and a third Demis Hassabis can emerge from a garage or a dorm room." What was born from that conviction was the open-source model 'Gemma.'
The name 'Gemma' comes from the Latin word for gem. It carries the meaning of releasing something precious into the world. Inside Google, however, voices of concern were loud.
The pushback asked why core technology developed with billions of dollars in investment should be given away for free, and whether it was just doing competitors a favor. Hassabis was looking at a bigger picture. With Meta seizing control of the open-source ecosystem through its Llama series, he made the strategic judgment that if Google clung only to a closed strategy, it would ultimately lose the support of developers. He also held firm to his conviction as a scholar that scientific discoveries grow exponentially in value when they are shared.
The release of Gemma 3 was the moment Hassabis's vision reached technical completion. Gemma 3 was not a scaled-down version of Gemini. DeepMind engineers performed a kind of magic, distilling the essence of a giant model's intelligence into a smaller vessel.
What made Gemma 3 remarkable was that it could run not on expensive server-grade GPUs but on an ordinary laptop or even a single GPU in a mobile device. It dramatically lowered the barrier to entry for AI research and application. A startup developer in Africa, a university student in India, a researcher at a small Korean company could all run AI on their local devices without an internet connection. Gemma 3 also possessed multimodal capabilities.
Beyond text, its ability to see and analyze images and process simple voice commands heralded an AI revolution on edge devices. A doctor providing medical care in a remote area with no internet could now take a photo of a patient's wound on a Gemma-equipped tablet and receive instant diagnostic support. At the Gemma launch event, Hassabis declared: "True democratization of technology happens when anyone, anywhere, can use the best tools without restrictions."
This aligned precisely with the motto he had set when founding DeepMind: "AI for Science, AI for Everyone." The ecosystem's response was explosive. Over 13 million developers worldwide downloaded Gemma and began building their own applications.
On GitHub and Hugging Face, derivative models fine-tuned for specific languages or adapted for specialized fields like law,
medicine, and coding sprouted up in enormous numbers. DeepMind researchers felt a sense of wonder watching Gemma being used in creative ways they had never imagined. Someone built a screen narrator for the visually impaired. Someone else built a translator to preserve endangered languages.
Hassabis described the phenomenon: "We threw a spark, and the entire world started a massive fireworks show." One development worth noting was the spread of 'Search AI Overviews' technology. Applied to Google's search engine, this feature uses Gemma's lightweight technology to quickly summarize and organize search results.
When this technology was released as open source, countless companies and websites were able to easily implement intelligent search within their own services. The result was a broad uplift in user experience across the web ecosystem. Users no longer had to click through link after link hunting for information; they could instantly get the key points organized by AI.
Gemma turned AI-powered search technology, which had nearly become the exclusive property of giant platform companies, into something resembling a universal web standard. The open-source strategy carried risks too. There were concerns that malicious users could modify the model to create hacking tools or generate fake news.
In response, Hassabis and the DeepMind team distributed a 'Responsible AI' toolkit alongside Gemma. It included safeguards to prevent the model from generating harmful content and tools to help developers use the model within ethical guidelines. Hassabis chose to trust the community's capacity for self-regulation and collective intelligence over technical lockdowns.
He reinterpreted the philosophy of Linus Torvalds for the AI era: "Keeping things closed may look safer, but publishing them transparently and watching over them together builds a safer and stronger system in the long run." Through the Gemma project, Hassabis showed himself to be more than a researcher or entrepreneur. He revealed himself as an architect of the technology ecosystem. Instead of a dystopian future where AI becomes the exclusive property of a few and concentrates power, he chose a future of 'fundamental abundance,' where everyone carries their own Gemma in their pocket and expands their intellectual capacity.
Thirteen million developers creating 13 million possibilities. That was the landscape Hassabis wanted to build by 'unlocking intelligence.' Gemma, small but hard like a gem, was another form of victory shining amid the race of giant models.
Project Astra and Project Mariner. May 2024, Shoreline Amphitheatre in Mountain View, California. When a video began playing on the Google I/O keynote stage screen, thousands in the audience held their breath. In the video, a user turns on a smartphone camera, walks around an office, and talks with an AI. "Where is the part of this speaker that produces sound?" the user asks, and the AI points precisely to the tweeter. The user shows the view outside the window and asks, "What neighborhood is that?" The AI recognizes the location and answers, "That's King's Cross, London." When the user can't remember where they left their glasses, the AI says, "You put them on that desk over there, next to the red apple, a little while ago." This was not a scene from the science fiction film "Her." It was a demonstration of 'Project Astra,' the ambitious creation prepared by DeepMind under Demis Hassabis.
Project Astra was the blueprint for the 'universal AI assistant' Hassabis had long dreamed of. He believed AI should not remain a text generator trapped inside a chat window. AI had to see, hear, and remember, just like a human.
Astra was equipped with the ability to understand and process video streams in real time. This was the result of breaking through a formidable technical barrier. The latency in converting visual information into linguistic information had to be reduced to near zero.
Hassabis drew fully on his background as a neuroscientist. He benchmarked the speed at which the human brain processes visual information and reacts, then designed Astra to carry on natural ping-pong conversations as if talking with a person. The core of Astra was 'Memory' and 'Context.'
Astra does not just recognize what is currently visible. It remembers what it saw moments ago and the full context of conversations it had with the user. This is equivalent to giving an AI a sense of time. Hassabis explained: "A true assistant is one that takes care of things without the owner having to explain everything each time.
Astra will be a second brain that remembers the places your eyes touched and the things you missed." This was a critical inflection point where AI evolved from a tool into a partner. Meanwhile, if Astra was an attempt to understand the physical world, 'Project Mariner' was an attempt to conquer the digital world.
Hassabis determined that the web browser, where we spend most of our day, was the stage where AI should be most active. Mariner is an autonomous agent that operates on top of the Chrome browser. Give it a single command like "Book flights and a hotel for my Tokyo business trip next week, compile a list of good restaurants, and put it all on my calendar," and Mariner
connects to travel sites on its own, compares prices, makes reservations, searches maps, and plans the route. The human just approves the results. Mariner's arrival is an event that rewrites the history of human-computer interaction (HCI). Until now, we had to click buttons and type with a mouse and keyboard ourselves.
Mariner is an 'Actionable AI' that moves the mouse and types on the keyboard by itself. To make this possible, the DeepMind team trained it on millions of hours of web browsing data. The AI learned to navigate complicated checkout processes on shopping sites and fill out finicky forms on government websites as well as a human, or faster and more accurately than one.
Hassabis compared it to "giving every person a skilled mariner to navigate the vast ocean of information that is the internet." This technical vision led to an unexpected major partnership: a collaboration with Apple.
The world was thrilled by the news that Google's Gemini would operate behind the scenes powering Siri on the iPhone. It was a pragmatic alliance that transcended Silicon Valley's long-standing rivalries. Apple needed a powerful cloud-based model to overcome the limits of on-device AI, and Google needed a channel to reach Gemini's potential audience of two billion iPhone users worldwide.
For Hassabis, this partnership was a decisive move toward popularizing AGI technology. Now people would hold up their iPhones, see the world through Astra's eyes, and travel the web with Mariner's hands. Hassabis predicted through Project Astra and Mariner that the future of AI would be an 'Ambient Presence.'
Like Jarvis in the film "Iron Man," invisible most of the time but appearing whenever and wherever needed to solve problems. That is the stage where technology has advanced so far that it becomes indistinguishable from magic. He always tells DeepMind engineers: "We are not building a product. We are designing the way people will live in the future." Astra and Mariner are the time machines heading toward that future.
Gemini 2.0
Kim Kyung-jin
Attorney · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.