Quote from: Scumby on September 21, 2014, 04:21:59 PM

I have only a basic understanding of Bitcoin's internals, so if there is nonsense here due to my own ignorance I apologize. I think you are doing a tour-de-force job of assimilating a wide cross-disciplinary breadth of ideas here, but I am concerned that the TexAI/Cyc ontology stuff is so alien to most developers that it could torpedo your adoption if you make it too front-and-center. The AI integration into the CPOS "stew" feels a bit forced; it's evidently a treasured ingredient in your own intellectual "refrigerator" and provides you continuity with your AI work, but you're going to have a hard enough time getting mindshare without having to also persuade engineers that symbolic programming with ontologies is worthwhile for cryptocurrencies. My .02.

Indeed.

For reason you mention, I have stripped out the work-in-progress, natural language dialog and face-recognition aspects of Texai from the code that I am committing to GitHub for TexaiCoin. When I talk with coin developers at the Hasher's United conference in Las Vegas next month, I will be talking about a multi-agent software system which can cooperate to mine coins with no effort and still be as secure as Bitcoin.

My approach to artificial general intelligence is to follow Alan Turing's advice back in 1950, which is to create a system capable of being taught skills and then to teach it. My implementation uses the robotic-inspired architecture described by James Albus, in which his hierarchical control network of software agents has been elaborated by me to model a human organization. For TexaiCoin, I have 27 simple agents which cooperate and will check each other to operate an altcoin network with no trusted third parties.

I could have implemented these agents for TexaiCoin using any one of several agent frameworks, but it was the security aspects that motivated me to repurpose my existing AI code. Concerned about the need to ensure Friendly AI in a distributed network operated by anyone downloading the code, I designed the AI code years ago to be secure against intruders and hackers. I incorporated X.509 certificates for that purpose. Currently, I plan to use tamper-evident logs and remote attestation to strengthen Texai against attackers.

I will be able to market TexaiCoin as having genuine artificial intelligence, thus helping this altcoin and its innovations to be distinguished from other thousand-plus altcoins. I do not plan to say much initially about the OpenCyc knowledge base that is part of each deployed container.

TexaiCoin ideally will become a platform upon which third-party developers can add vetted skilled agents, and receive payment in TexaiCoin for the benefits gained by the collective infrastructure, or from the proceeds of services for sale, e.g. an API. I will start with the document timestamp service as a demonstration.

I will add a high level goal to the system to pursue self-improvement, e.g. to increase revenue and coin value via tasks performed by paid developers. Humans will be able to perform tasks specified in the same manner as for software agents, but with a secure web form or secure email as the communication channel. I think that the system paying one person to perform the task, and paying two non-affiliated others to check the work should be safe.