Dna2agi is based on the following ideas:
- Intelligence exists in many forms in biological life.
- A software neural network is a construct loosely modeled after a biological brain to produce intelligent behavior.
- Intelligent behaviors existing outside of the brain in biology can also be modeled via software to produce problem-solving capabilities.
- This includes biological systems that navigate transcriptional, metabolic, morphological, and organismal problem spaces.
- The minimum context for general intelligence in biological organisms to operate may require the span of intelligent systems working in concert via hierarchical and modular interfaces, not just one surface layer like the brain.
Our working theory is that a probable path to Artificial General Intelligence (AGI) is a holistic yet simplified modeling of several biological layers and the interactions between them. The full stack includes the networks and pathways involved in genomic, transcriptional, physiological, morphological, and organismal layers.
We aim for a strong reductive bias towards capturing only the essential primitives, structures, functions, and interfaces relevant to intelligent behavior. High fidelity is not a goal; rather, we are searching for minimum viable models for general intelligence. Evolution may have already done most of the selective legwork for us.
We'll develop a human interface to fluidly work with these models. We believe this is essential to success.
This is an audacious and perhaps absurd path towards AGI, yet it might prove the most reliable. Just ask a mouse or a dolphin! 🐭 🐬 🌍 🌌