# SIML Demo: Optimizing AI for Energy and Efficiency

An early embodied-agent demonstration exploring adaptive memory and prediction in a simulated world.

Date: April 15, 2025
Type: Historical prototype
Status: Historical
Credit: Author and developer
Authors: Richard Everts

## A world for the agent

The original SIML demonstration places agents in a voxel environment and explores how internal memory changes alongside interaction with that environment. It is an artifact of the program's early work on adaptation and surprise.

Its accompanying page describes a compact architecture and links to code and a demonstration video. The point of keeping the prototype visible is to make the development process inspectable, including the assumptions used at the time.

## Historical status

Deep SIML labels this prototype historical. Its original artificial-life terminology and broad performance language should not be read as the lab's current claims or as proof of general intelligence.

The later research archive provides more narrowly framed studies with their own methods and limitations.

## Sources
- [Original prototype](https://www.siml.life/research/siml-demo)