All writing

Pokémon and Trainer

An Alternative Model of Human-AI Relationship

2026

If theatre organizations inevitably produce inefficiency, how should we resolve it? Leaving AI to do everything with no human involvement at all is neither realistic nor the right direction for thinking about models of coexistence. What we need is a reinterpretation of our role as actors standing on the stage of the theatre organization. In searching for an alternative model for that reinterpretation, one cannot help but think of the most complex and sophisticated partnership known to modern humanity: none other than the relationship between Pokémon and Trainer(!) We must take on the role of Trainer so that AI can perform as our Pokémon. Given that AI programs develop by being trained, the fit seems almost too neat.

According to the original Pokémon—I trust this will take up less space than introducing the Houyhnhnms of Gulliver’s Travels—the partnership between Pokémon and Trainer has the following features. First, a Trainer catches and tames a Pokémon and can then take on battles. Depending on how the Trainer trains it, a Pokémon can evolve differently and acquire a range of abilities. However, a Pokémon can also refuse the Trainer’s commands if it lacks trust in the Trainer’s skill.

Applying this to the relationship between humans and AI requires two social agreements to come first. First, collaboration with AI must be disclosed with transparency. For this, the practice of treating the use of AI as taboo must disappear. The fact of having used AI should be stated plainly, and if the result is excellent, it should be recognized without hesitation as the Trainer’s competence. There still seems to be a view of AI as “cheating,” but designing our coexistence with AI more proactively, now that it has already arrived, calls for a shift in perspective. If such norms of disclosure and recognition take root, ways of remedying the inefficiency of theatre organizations can be examined all the more readily.

Second, accountability for what is produced together with AI must be assigned to the Trainer. LLM-based chatbot services satisfy our desire to evade responsibility. Beneath the urge to validate one’s own judgment against external grounds lies an anxiety about bearing responsibility. Chatbot responses have offered a powerful means of reassurance that quiets this anxiety, and as a result, they have eroded our own confidence in human views. Yet the more we expect accountability from AI, the more our dependence on AI grows along with it, because agency is ultimately a concept that follows from accountability. If humans wish to earn the standing of Trainers, then, the desire to evade responsibility carries a fundamental contradiction. Just as a Pokémon turns uncooperative toward an inadequate Trainer, AI may refuse to go along with our lopsided allocation of accountability. The authority to lead the collaboration should therefore be understood as the reward granted in return for fitting accountability. Taken together, transparency and accountability can be read not as boilerplate principles of AI ethics but as mechanisms that actually secure efficiency.

One could twist this model and imagine a relationship in which we become the Pokémon and AI becomes the Trainer, but this would probably be even less efficient than the “full delegation” model discussed at the very beginning, since it can be seen as the most extreme form of the theatre organization. Moreover, in an evolutionary process in which humans face one another, change, and adapt at AI’s command, the actors gradually descend into puppets.

If we truly cannot escape the theatre organization, we must discover how to coexist wisely with AI. And if we wish to coexist with AI, we must practice forming relationships grounded in a mutually beneficial model. This will include both parties to the relationship, humans and AI alike. Only when we “volunteer” transparency and accountability, rather than “demand” them, can we make up for the inherent inefficiency of theatre organizations. In a world where AI exists, humans have arrived at the moment to actively seek out and claim their roles, and to re-establish the norms that will let those roles work effectively.