Traditional storytelling, whether single-threaded or interactive, is still based on “events” as the basic unit
of narrative, that is, what happens and what happens. Interactive digital entertainment is nothing more
than giving the user the freedom to choose from a given two or three options for different events to
happen next, and the entire narrative is still based on a pre-defined path forward.
Figure 2: Traditional Storytelling: Single-threaded Narrative
Figure 3: Traditional Storytelling: Interactive Narrative
Quantum Hunter GameAI Engine called Heisenberg Engine is completely different from traditional
narrative, and the plot of each game user is filled with Uncertainty, as if Heisenberg’s Uncertainty Principle
in quantum mechanics.
The Heisenberg Engine is a creative system that can use a core AI algorithm to extract the script and
generate the main storyline, opening up almost infinite narrative possibilities and story structure, while
decoding the meaning of every word and translating it into 3D rendered animation.
We replace “events” with an infinite number of black boxes with clearly defined entrances and exits. In
each slice, the beginning and end (one or more) are determined, but how the player gets from the
beginning to the end each time is uncertain. This path can only be determined when the player is
constantly interacting with other players, NPCs, and the environment in the game world that are moving
the story forward by reacting dynamically and in real time to models trained in deep reinforcement
So the key to a truly interactive narrative is to shift the center of the narrative from the story itself to all
of the possible participants in the story, and to let the logic of all of the possible participants work
together to drive and connect different narrative possibilities. Within the scene are characters, including
players and NPCs, as well as objects in the environment that can be interacted with. A person has a
character, a state, a set of actions; Objects have physical Settings, including orientation, size, shape,
color, etc., as well as states and supported actions.
Once a character has a state and setting, it has an effect on the set of possible actions. With the input,
setting and shutdown conditions, the simulation environment and deep reinforcement learning model
can be used to explore the behavior strategy of each character (including players and NPCs) in the closed
scene, and learn a reasonable decision-making model that is in line with the target strategy. At the same
time, the strategies explored in one scenario (including the strategy set of characters and objects) can be
dismantled and integrated so that such strategies can be reused and evolved in subsequent scenarios.
Quantum Hunter is dedicated to enabling intelligent behavior for the avatars in the game, allowing the
player to drive the infinite progression of the story and plot, and exploring more open-ended endings.
It’s definitely a very interesting and challenging job.