felgate

Advocate High Court SA. Developer of Science-based Rule of Law on how it arises, its mechanics & effects. Motivator of Supra-international ROL-based body

[felgate @felgate]

@AnthropicPolicy governs dangerous AI use cases. It is not governing the most common one: AI drafting and operationalizing instruments of state power inside government. The gap is closeable. One ROL due diligence clause. No legislation required. https://t.co/ha0dcaES9f

[felgate @felgate]

@sama So what's your plan? How u going to make government govern AI when they don't know how to? My blog has some answers – Science of Rule Of Law "The Contractual Standard That Would Make AI Governance Workable" https://t.co/x8klSHNY6x

[felgate @felgate]

When it's necessary to kill them" — Palantir CEO, Feb 2025 earnings call. That's an exercise of state power. It needs ROL compliance. A published standard now makes that contractually enforceable. #AIGovernance #RuleOfLaw #Palantir https://t.co/7M6Zro6EhA

[felgate @felgate]

@elonmusk Especially to someone who fibs

[felgate @felgate]

The Rule of law hasn't failed. Its understanding has.

After 30 years fused with democracy. Each defining the other, neither independently testable. Governance breaks down. Now measurable

https://t.co/G1wyzOuWYl

#RuleOfLaw #Governance #EarlyWarning #OpenAccess

[felgate @felgate]

MEASURING RULE OF LAW THROUGH DOCUMENTARY EVIDENCE A Seven-Category Methodology Applied to South African Government Gazette Data, 2008-2024 https://t.co/osQ04HYAFW

[felgate @felgate]

Every generation has its governance breakthrough. Each hit the same wall: undefined terms no operational standard no enforcement

The Rule of Law is different: 20yrs data 7 testable categories & the oldest accountability governance is finally precise

https://t.co/5tneABufbC

[felgate @felgate]

AI Moratorium Debate Asks the Wrong Question: The real issue is de facto uncontrolled application of AI in gov & the real question is whether existing, deployed AI systems are meeting Rule of Law standards https://t.co/7M6Zro6EhA

[felgate @felgate]

Any methodologists around familiar with content analysis (inter-rater reliability design, codebook validatio), Kappa statistics (Cohen's / Fleiss, weighted Kappa), Calibration validity (specifically whether external criterion measure can address construction overlap concern)?

[felgate @felgate]

@isareksopuro We've now got to the State Commons :)

[felgate @felgate]

@isareksopuro And I've spent 3 months getting Claude to understand the Rule of law, as the framework within which to govern - we've got up to the Jurisdictional Gap between states. We should get together :)

[felgate @felgate]

@isareksopuro And I've spent 3 months getting Claude to understand the Rule of law, as the framework within which to govern - we've got up to the Jurisdictional Gap between states. We should get together :)

[felgate @felgate]

@robert_zubrin @lucyshow11 10

[felgate @felgate]

AI performing governance functions needs coordination architecture—not new industrial policy or ethical principles.

Component separation. RRL verification. Gatekeeper architecture.

Foundation before dreamscaping.

https://t.co/fWdlMhwgJ9

[felgate @felgate]

On AI deployment approaches:

- OpenAI: "Safety nets when bridge fails"
- Anthropic: "Bridge with good values"
- Grok: "Less woke bridge"
- SROL: "Missing structural supports"

None specify coordination architecture.

https://t.co/fWdlMhwgJ9

[felgate @felgate]

Why share openly?

If it works beyond AI, research community should have it.
If it doesn't generalize, that's also useful data.
Either way, better to test than hoard.

[felgate @felgate]

Sharing a methodology paper that emerged from governance research with AI systems. May interest phenomenologists & cognitive scientists.

Released freely for testing - curious if it works with human subjects.

https://t.co/EXDlziMU7d

[felgate @felgate]

Released under CC0 (public domain). Test it, adapt it, critique it freely.

If anyone tests with human subjects and finds interesting results (confirming or disconfirming), I'd be curious to hear.

[felgate @felgate]

@sciencegirl Chiaroscuro