self-evolving Internet-deployed intelligence

[synctext]

Goal: create an intelligent entity which can autonomously expand it's core self.

In prior work we developed real systems with full autonomy and special properties. We created self-replicating entities in the past years based on the work of in total 16 students. We demonstrated self-compiling smartphone apps with viral spreading. Our decentral market is fully operational and has active asset trading.

The next step in this deliberately scarry and inspirational journey is a self-evolving entity. A real implementation with real autonomy and with full self-governance. Our self-evolving entity lives in our token economy, consists of Bitcoin-paid VPS servers, and earns a living by offering privacy services. The evolutionary mechanism consists of a public "Python DNA repository" through which mutations spread and from which new-born agents select mutations. The Python DNA repository is a distributed storage facility for Python software and features tamper-proof tracking of the historical performance of each mutation using Trustchain.

Our DNA repo offers a bounty for mutations which increase survival and general fitness. The competition for the scarce token resource is the driving force for survival in our self-evolving ecosystem. Each mutation is simply a Bittorrent swarm filled with Python source code in GIT diff format, identified with a single hash and it can be possibly combined with other mutations forming a mutation set. Each intelligent entity follows a regime of self-imposed taxation. Taxes go to the owners of the mutation creators, identified by their performance tracking trustchain genesis block. Intelligent processes, random processes, and humans alike can contribute mutations to the DNA repo. The self-evolving and self-replicating entities have a strong preference to re-use and randomly re-combine strains of DNA with proven multi-generational survival capability for their offspring. Self-contained unit tests are used to test the liveness property and compatibility of mutations.

This is a scientific exercise into "economic engineering". Is this ecosystem sustainable and can it overcome the tragedy of the commons problem? It would represent an advanced circular and sustainable micro-economy. Based on solid game theory and market principles.

Our distant goal is to have forrests, cars and entities in general which own themselves (thnx for this idea Dimitri!). Perhaps we can then keep humanity within the ecological constraints of our planet. Managing finite resources is too important to leave to politicians.

[ichorid]

The hardest problem here is to define the "genome API" which determines how an AI entity will interact with the world. Another great problem is malicious human programmers putting in long-term "code bombs" into proposed genes. Humans to autonomous entities are like viruses to bacteriae, with their relationships mostly, but not entirely parasitic (bacteriae use viruses as vectors for transmitting genetic information). On the other hand, bacteriae have a special defense subsystem (CRISPR) against viruses...

So, there should be no direct input by human programmers into AI DNA.

[synctext]

Related work: AI and economic impact https://thegradient.pub/the-economics-of-ai-today/
Quotes:
The arrival of AI systems able to perform some of these tasks impacts on demand for labor, the share of income that goes to it (or to capital), and inequality
It looks like modern technologies such as AI increase demand for high skilled jobs that complement AI and low-skill jobs that are difficult to replace with AI, leading to polarization in the labor market.
Prasanna Tambe and co-authors also use LinkedIn data to estimate the value of intangible investments related to AI, finding that it is concentrated in a small group of "superstar firms"

Conclusion: Think Internet, not Skynet