Applications and the Reference Databases
Every activity that depends on people seeing and responding works with human behaviour every day: retail and the shop floor, transport and the movement of crowds, public health and the routines of daily life, education, safety, the workplace, politics, advertising and media.
Applications
Scene Intelligence gives each of them one equation, finite catalogues and a frontier AI that reads the scene as the participant reads it.
Virtual insertion
A further use is the ability to modify what can be seen without disturbing the original creation, enabling changes that can influence the viewer’s thoughts and feelings.
Permission
Every insertion needs contractual permission from the content owner, and the art, narrative and creation must be respected.
The billboard concept
The billboard concept uses this: a virtual billboard is inserted into a scene and its content is then changed repeatedly, as the advertisement on a real billboard changes over time, so that the scene becomes a canvas that can host different inserted advertisements in turn.
From insertion to understanding
The question moved from “What is in this scene?” to “What is happening and what will happen next, given what’s here and what we know of human nature?”
From seeing to shaping
Knowing U, a creator can change the scene to move U in a chosen direction, and ask “If I change this element, will the outcome (U) change?”
Levels of scene understanding
Perceives objects
Recognises basic scene types
Understands interactions and can narrate what is happening
Anticipates what will happen next in the scene
Actively shapes scenes to achieve chosen outcomes
Equation Two
Equation One is an input to Equation Two: U enters Cb directly, so no change can be assessed without first reading the scene.
Reference databases
Resolving a scene needs reference databases to resolve it against.
Computing views
Computing views is where scene intelligence is applied at scale.
The Compute That Makes This Possible
The equation is computed on hardware.