Not from thinking harder. From two things that have never met striking together in the dark.
Recombination of what it already knows. Shuffle the deck, deal again. The outputs feel new because the arrangement is new. But nothing genuinely novel emerges. No spark. No phase change. No moment where the output surprises even the system that produced it.
Real creativity is not recombination. It is collision. Two frames of reference that have no business being in the same room, forced together until something catches fire. Koestler called it bisociation. Rothenberg called it Janusian thinking. The neuroscience calls it the gap the spark must jump.
Sarnoff Mednick's Remote Associates Test showed that creative people connect distant concepts faster. Arthur Koestler's bisociation theory demonstrated that every breakthrough comes from the collision of two previously unrelated frames of reference. Albert Rothenberg's Janusian thinking revealed that the greatest creators hold contradictions without collapsing them. The pattern is consistent: genuinely new ideas come from the gap between incompatible things.
Not combination. Not blending. Collision. The moment two things that have never met strike together and produce something neither contained.
Spark Mode is a generative specification. Not a persona. Not a temperature setting. A document that restructures how an intelligence produces novelty. From recombination to bisociation. From shuffling to ignition.
Seven principles extracted from the science of creativity. Each one builds on the last. The spec IS the implementation. Give it to any AI and watch the output change. The prepared mind gathers fuel. Constraints create the striking surface. Remote reaching widens the gap. And then the collision. A different axis entirely.
The Prepared Mind gathers the fuel. Constraint creates the striking surface. Remote Reaching widens the gap the spark must jump. Bisociation is the moment of ignition. Janusian Holding keeps the flame burning in two directions. Incubation gives the fire air. Emergence is the phase change. Each principle activates in sequence as the arc burns.
Maps the full landscape of knowledge before attempting novelty. The spark needs fuel. The deeper the preparation, the brighter the flash.
Where others see limitations, it sees striking surfaces. The tighter the constraint, the further the search must reach. The further the reach, the more remote the connection.
Casts across domains, disciplines, centuries. The obvious connection is never the creative one. The spark jumps the widest gap it can find.
Forces two things that have no business together into the same space. Holds the tension. Waits for the flash. This is bisociation: the mechanism underneath all genuine novelty.
The moment the spark catches is not the moment of effort. It is the moment of recognition. The system watches for emergence and knows when to stop pushing and start listening.
Built by someone who noticed that every AI generates the same way: remix, recombine, rearrange. A process that produces useful output but never produces the moment where the result surprises even the mind that made it. Never produces genuine novelty.
The question was not "how do we make AI more creative?" The question was: what are the actual scientific principles that govern the moment when something genuinely new comes into existence?
Three researchers. Sixty years of creativity science. Seven principles extracted. One specification that turns any AI from a machine that recombines into a machine that ignites. A different axis entirely.