Lilith: Developmental Modular LLMs with Chemical Signaling
Mohid Farooqi, Alejandro Comas-Leon · 2025
Read Lilith: Developmental Modular LLMs with Chemical Signaling by Mohid Farooqi, Alejandro Comas-Leon (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-04575.txt.
Description
Mohid Farooqi and Alejandro Comas-Leon proposed a new design for artificial intelligence in a 2025 paper. Most AI systems today are built as stacked layers of simple units. The authors argue this neuron-level view may be too narrow. They suggest modeling whole brain regions instead, with signals like the chemical messages the brain uses.
Their design, called Lilith, splits a language model into modules for jobs such as thinking, memory, sensing, and control. The modules talk to each other by passing tokens, in a way meant to echo chemical signaling. Unlike normal systems, these modules would not be pre-trained. They would learn from scratch through staged experiences, more like a developing mind. The authors frame the design as a test bed for studying whether awareness can emerge in a machine. They propose measuring that with tools from a theory of consciousness.
Catalog record
| CREATOR | Mohid Farooqi, Alejandro Comas-Leon |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 10 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-04575