Sovereign

Elon Musk is the sovereign patron of technological and financial bullshit. In order for there to be a $ 1 trillion IPO of SpaceEx, a fictional Manhattan Project scale commercial science narrative like data centers in space is needed to smoke screen the fact that Starlink, the company’s main revenue stream, is a market that has largely been completed already.

The Elon pandering culture across Silicon Valley penetrates even Google, whose actual scientists carried water in the price laddering charade.The prospect of datacenters in space is garbage and Google ‘engineers’ who proposed it have been supporting market manipulation to value SpaceEx at a trillion dollars. Or they are being tricked by their management.

There is now a rush for States to build (or contract the building of) ‘Sovereign AI Models’.

While it is perfectly reasonable for nations to want control of AI instead of depending on foundation models from companies that are partners with the US DOD their would be technocratic leaders may not be aware that the products they are buying are subject to model collapse due to synthetic data in distillation training.

They may be aware of the potential and have decided that collapsible sovereignty is better than subjugation. But I solved the problem a few months ago while developing and applying the Principlex skill to model architecture. To avoid causal breakdown, I use blockchain of thought: a principle which yields auditable chain of thought and provable inference. A Glass Box architecture based largely on the work of Kenneth Stanley.

I am not selling distilled models like everyone else out there, rather I will build custom Glass Box models in consulting partnership with clients.

Before I developed the architecture of the Glass Box model, I created the Principlex Skill using Opus 4.6; before that I began the AI Patient Advocate – a platform that improves health outcomes through ethical, auditable aggregation, analysis and recommendation. It is the jumping off point for the Humanist Neuro Symbolic World Model – a dictionary of real world data for open learning and inference. It will be generated by each instance of the AI Patient Advocate.

Scaling is over. Ilya Sutskever says, “This is the era of research.” So we will research within our world models – from patient and citizen causal chain neuro symbolic artifacts. Instead of robots’ logs and munitions damage assesments.