Can AI Surpass the Pokémon Trading Card Game? A New Challenge for AI in 'Luck' and 'Inspiration'
Chapter 1: AI Takes on Another New Game
It won at chess, it won at shogi, and it won at Go.
I think you have heard the news many times over the past few years that AI has beaten humans. And in 2026, AI stood on a new stage of challenge. The opponent is the 'Pokémon Trading Card Game,' commonly known as Pokémon TCG.
A large-scale tournament was launched in which AI researchers from all over the world are participating.
It is hosted by The Pokémon Company and Kaggle (a global platform where AI skill-testing contests are held). Furthermore, HEROZ, Inc., which created the shogi AI 'Ponanza,' and the Matsuo Institute, known for AI research, are listed as co-hosts, and Google and NVIDIA are also cooperating, making it a very serious tournament. AI agents will take on the Pokémon TCG, where you choose and build a deck of 60 cards and aim for victory through tactical exchanges with your opponent.
Ultimately, this is not just a test of whether AI can win a game. It is a grand experiment to see how far AI can imitate human 'luck,' 'inspiration,' and 'tactics.'
Chapter 2: What Makes the Pokémon TCG Different from Chess, Shogi, and Go?
Here, I will organize one important difference.
Chess, shogi, and Go are all what are called 'perfect information games.' In short: they are games where the state of the board is fully visible and there is no hidden information. Both the opponent's pieces and your own pieces are fully exposed on the board. That is why AI has surpassed humans through brute force, 'reading dozens of moves ahead.'
However, the Pokémon TCG is different.
The opponent's hand is not visible. Which cards you draw from your deck is a matter of luck. Furthermore, the Pokémon TCG has a mechanism called 'Prize cards.' This is a system where you place 6 cards from your deck face down at the start of the match and draw them as a reward every time you knock out an opponent's Pokémon. In other words, even you yourself do not know what you will draw until the very end.
Not only is the opponent's information invisible, but you must continue to calculate the expected value for winning while the contents of your own resources remain uncertain.
Among the combinations of cards, there are times when a single card that looks plain and inconspicuous at first glance is actually the key to the entire deck.
In other words, what AI faces in the Pokémon TCG is not 'depth of reading,' but a completely different kind of wall: 'how to make decisions amidst uncertainty.'
Until now, AI has demonstrated its power in worlds where the board is fully visible. What it is about to take on is how to deal with a more ambiguous world where luck and inspiration are mixed in.
Chapter 3: Can 'Unexpected' Things Happen to AI? What the Movie 'AWAKE' Teaches Us
What I want you to remember here is the movie 'AWAKE,' released in 2020.
This movie, starring Ryo Yoshizawa, is based on the 'Den-O Shogi Tournament' that actually took place in 2015, a match between a professional shogi player and computer shogi software.
The model for the AI 'AWAKE' that appears in the film is the real-life shogi software 'AWAKE.' It was developed by Ryoichi Kose, a person who had originally entered the 'Shoreikai,' a training institution for shogi, aiming to become a professional shogi player.
In the actual Den-O Tournament, because the professional shogi player used a 'certain sequence of moves' that amateur shogi fans had discovered, AWAKE was defeated in the blink of an eye, in just 21 moves.
AWAKE was not a weak AI; in fact, it was one of the strongest pieces of software at the time. Even so, it could not demonstrate its true potential against an 'unexpected move' that it had not learned.
This remains a persistent weakness for AI even today.
Even if an AI is strong within the scope of the data it has learned, it can suddenly become weak or make strange judgments when it encounters a situation it never anticipated.
Technically, this is a phenomenon known as 'out-of-distribution'; simply put, it is a phenomenon where an AI's judgment suddenly becomes unreliable the moment it encounters a situation it has never experienced before.
Since it hit this wall even in the world of Shogi, where the rules are fixed, don't you think it's worth watching what kind of 'unexpected' situations an AI will encounter in the Pokémon TCG, where there are many types of cards and luck is also involved?
Chapter 4: What is Unique About This Pokémon TCG Tournament?
There are two divisions in this AI Battle Challenge.
One is the 'Simulation Division'.
AIs repeatedly play automated matches against each other, and the ranking is determined by their win rate.
The other is the 'Strategy Division'.
This is the interesting part: AIs are required to explain their 'way of thinking' to humans in a report, detailing why they built that deck and why they chose that specific move.
In other words, this tournament asks AI not only to 'win' but also to 'be able to explain'.
It is attempting to make AI explain in words the intuitive judgments that humans make in Shogi or Mahjong, such as 'I just have a feeling this is the right move'.
How far can this be done, and if it cannot, what will be the barrier?
I think this is a quite challenging attempt even as AI research.
Chapter 5: What Ryoichi Kose's Subsequent Path Teaches Us
Many people who watched the movie 'AWAKE' might be wondering, 'What happened to Kose after that?'
In fact, Kose is currently active in a field completely different from Shogi.
He serves as the representative of a company that researches simulation technology for MRI (Magnetic Resonance Imaging, a device often used for hospital examinations).
The ability to 'design complex calculations' that he cultivated through the development of Shogi software is being utilized in another cutting-edge field: medical technology.
I think this is a very important implication.
The experience of seriously devoting oneself to creating Shogi software did not stop in the world of Shogi, but led to a career in a completely different field.
Among the young developers taking on this Pokémon TCG AI challenge, there might be people who, in the same way, 'get hooked on the world of AI because of the Pokémon TCG and go on to be active in a completely different field in the future'.
Chapter 6: A More Interesting Question Than Whether It Will Take Our Jobs
Recently, news about AI has tended to be filled with dark topics about how 'jobs might be taken away'.
But this Pokémon TCG challenge shows us AI from a different angle.
It is a question that purely tickles intellectual curiosity: not whether AI will beat humans or lose, but 'to what extent can AI understand the fun of luck, inspiration, and tactical maneuvering that humans possess?'
Because it is a pure experiment where no one is troubled even if it loses, we can safely look forward to the results.
The results of the first round are yet to come, and the progress of the second round is scheduled to be distributed on YouTube after the end of 2026.
The conclusion is still a long way off, but please pay attention to what kind of moves AI will make in a paper card game, a pastime that has been established for decades solely through human-to-human tactical maneuvering.
If you read this article and became curious about 'what kind of judgments I am making', there is one thing I would like you to try the next time you play Pokémon TCG, Shogi, or Othello with someone.
Whether you won or lost the match, try leaving a note with just one sentence about 'why you chose that move in that situation'.
You should be able to experience for yourself the 'explainable judgment' that AI is currently tackling.
Whether this article sparks your interest in the Pokémon TCG or in AI development, I would be happy either way.
And I look forward to the day when someone who, like Ryoichi Kose, uses the experience of having been truly dedicated to something as fuel to succeed in a new field, emerges from this tournament.
This discussion surrounding AI and human intelligence is also covered from a different angle in a previous article, Is 'Mythos is Scary' True? The Government Has Acted. Your Company Is No Longer Just Watching a 'Fire on the Opposite Shore'. It is connected by the themes of 'How much can we trust AI?' and 'What are the judgments that only humans can make?', so please read it as well.
If you have any thoughts on this AI x Pokémon TCG topic, or if you have seen the movie 'AWAKE,' please share them with me.