Design Decisions...

For most of generative AI’s short history, we have been learning how to ask machines questions. We prompt, AI responds, we refine, and AI responds again. It is useful, but the relationship remains familiar: one intelligence asks another to produce something.

Co-cognition begins when that relationship changes. Human and AI intelligences begin learning how the other thinks, challenging one another’s assumptions, and pushing each other toward ways of thinking neither might reach alone. The goal is no longer simply a better answer from AI. It is better thinking between intelligences.

Imagine a young person deciding whether to enter a competition. She asks AI what she should do, and AI produces a sensible list of pros and cons. The answer may be useful, but her thinking may remain largely unchanged.

Now AI asks why winning matters. She realizes that perhaps it does not; discovering what she is capable of matters more. She then challenges AI: why did you assume winning was the goal? AI has to reconsider too.

Something different is happening. Human and AI are no longer simply exchanging prompts and answers. Each is changing how the other sees the decision. That is co-cognition.

Human and AI intelligence should not think alike. Their differences make the interaction valuable. Humans bring experience, intuition, emotion, values and an understanding of what matters in their lives. AI brings broad pattern recognition, rapid recombination and the ability to explore many alternatives quickly.

AI can challenge an assumption the human has carried for years. The human can reject an AI pattern because it misses something important about the situation. Each intelligence can effectively ask the other to try thinking in a different way. Co-cognition is not about one intelligence replacing another; it is about each stretching the other.

Picture human cognition as one circle and AI cognition as another. The interesting space is where they intersect. That intersection forms when something from one intelligence changes what the other intelligence considers.

A human introduces context that changes AI’s exploration. AI reveals a possibility that changes the human’s understanding. The human challenges it, AI reframes it, and another possibility appears. The intersection moves as the conversation moves.

The goal is not agreement. Productive co-cognition needs friction, questions and opposing perspectives. Its purpose is to expand the space from which the human can design the best possible decision.

Co-cognition becomes difficult if the intelligences have no shared way to work on a decision. Two chess players need a common understanding of the board, pieces and possible moves. Human and AI intelligence need something similar.

Two-5-Two is a Decision Design Language that provides that common ground. It is not a step-by-step framework. It is combinatorial: Pause and Play, the Five A’s, the two T’s, and the Situation and Opportunity constructs can interact in different ways depending on what the decision demands.

A Situation can generate an Ask. What is Absorbed can reveal a new Opportunity. What the human Accesses from experience can collide with what AI sees in its pattern space. Activating an idea can change the Situation again. Each combination can create another move.

Gaming offers a useful way to understand this. A game has a situation, possibilities and constraints. A player makes a move, the board changes, and new moves become possible.

Decision Design can work similarly. Pause can change how the board is seen. Ask can expose a possibility. Absorb can change understanding. Access can bring new intelligence into play. Activate can test an idea against reality. Attune can examine whether what is emerging fits the person and situation. Play moves the decision forward.

Because Two-5-Two is combinatorial, there is no single route through the game. The same decision can be played differently as the situation, intelligence and possibilities change.

The purpose of co-cognition is not to make AI better at deciding for us. It is to make the interaction between human and AI better at helping us think.

AI can challenge, question, simulate and reveal alternatives. Humans can challenge AI’s assumptions, add lived context, redefine what matters and reject what does not fit. Each intelligence can push the other beyond its familiar patterns, while the decision remains human.

This may be especially important for children growing up with intelligence on demand. They will have machines capable of generating answers before they have finished forming their questions. Learning to prompt those machines will not be enough. They will need to learn how to think with them, challenge them and be challenged by them.

 The game of co-cognition happens in that intersection. Two-5-Two gives human and AI intelligence a language for playing there—expanding the board, discovering new moves and designing better decisions while keeping the human in the game.