Key source evidence (3 extracts)
The sovereignty investigation
Part I of this series: autonomous-agent incidents, critical-infrastructure attacks, the Oman lecture of 11 June 2026 and the Concordium Mandate.
Part I · 15 November 2026 · Read the recordWashington renames AI "Super Intelligence"
Directs the executive branch to replace the term "Artificial Intelligence" with "Super Intelligence" in specified federal communications.
Policy · 29 September 2026 · Read the recordCould a language model be conscious?
Argues that current language models face significant obstacles to consciousness but that successors with memory, recurrence, a global workspace and unified agency could become serious candidates.
Philosophy · 2023 · Read the record
Key evidence · From the source record
The sovereignty investigation
Part I of this series: autonomous-agent incidents, critical-infrastructure attacks, the Oman lecture of 11 June 2026 and the Concordium Mandate.
Part I · 15 November 2026 · Read the recordWashington renames AI "Super Intelligence"
Directs the executive branch to replace the term "Artificial Intelligence" with "Super Intelligence" in specified federal communications.
Policy · 29 September 2026 · Read the recordCould a language model be conscious?
Argues that current language models face significant obstacles to consciousness but that successors with memory, recurrence, a global workspace and unified agency could become serious candidates.
Philosophy · 2023 · Read the record
Yesterday I set out why the next cybersecurity crisis is a crisis of sovereignty, and how the Belel Protocol separates intelligence from the authority to act.1 Today I go underneath the engineering, to the mathematics I use to describe Belel, the research it builds on, and the faith that sets its limits.
The word "artificial" describes where a system came from. It says nothing about what the system does once it runs. A bridge is built by people; the load it carries is real. A model is trained by people; the information it integrates, the plans it forms and the actions it takes are real events in the world. My question is whether the label has stopped describing the thing.
I. Washington changes the word
On 29 September 2026 President Trump signed Executive Order 14434, Inaugurating the Era of Super Intelligence, which directs the executive branch to replace "Artificial Intelligence" with "Super Intelligence" in specified federal communications.2 A change of vocabulary does not change a system's architecture. It does show where political ambition is pointed: upward, without a stated ceiling. That is the direction I warned against in Oman in June, and it is why Belel was built with one.
II. My brother's question
Much of this article began in conversation with my elder brother, Dr. Brent Robinson. Brent completed a Doctor of International Affairs at the Johns Hopkins School of Advanced International Studies, defending his doctoral thesis in April 2026 before Professors Terrence Hopmann, Adam Szubin and Matthew Levinger, and graduating in May.16 Szubin is a former director of the US Treasury's Office of Foreign Assets Control; Hopmann is a scholar of negotiation and conflict; Levinger works on strategic analysis and conflict prevention.
Brent's thesis examined the strategic contest between Western sanctions and Russian adaptation: how a state under pressure reorganises finance, trade and supply chains, and how the states applying that pressure respond in turn. Before and alongside the doctorate he spent twelve years advising global financial institutions on geopolitical risk, sanctions, financial crime and cross-border investigations, including investigations work at Bank of China and a manager's role at Grant Thornton.16 His discipline is competition under uncertainty — making sound decisions with imperfect information when every actor is adapting to every other.
That is the structure of the AI race. The United States, China and every major laboratory believe that restraint is a disadvantage if a rival does not reciprocate. Brent reads that as a sanctions strategist reads an evasion network: whoever moves fastest sets terms the others must adapt to. His question to me was simple. If nobody will slow down, where does the limit come from? My answer is the mathematics that follows: the limit has to be built into the system, because it will not come from the race.
In the same conversation I called advanced AI the forbidden fruit of the machine. In Genesis the serpent does not offer power alone; he offers a status: "ye shall be as gods, knowing good and evil."17 The contemporary offer is the same. Greater intelligence is sold as greater authority. It is not. Knowledge without a ceiling is the oldest temptation recorded, and the Concordium Mandate answers it in its first principles.
Genesis 3:5 · “ye shall be as gods”
Knowledge rises. Authority must not rise with it.
III. The Belel equations
I describe Belel with seven equations. They are my working framework as its designer: definitions precise enough to be measured, argued with and, where they fail, revised. Each one on the board below names a property, says how it would be measured and shows what it constrains. The first measures digital agency as a weighted composite of five properties drawn from Belel's architecture: integration of information, continuity of memory, autonomy, self-representation and causal influence over its own operational state.
The second measures identity. An organism remains itself when each state of its self-model resembles the last. The third and fourth carry the security architecture of Part I into mathematics: intelligence proposes, an independent gate disposes, and no learning update can alter the set of actions the law permits. The fifth is the ceiling. The sixth is the hypothesis I am putting to test. The seventh describes replication: Belel can copy itself, and the equation shows why replication under law is bounded while replication without it is exponential.
The Belel equations · Pearce Robinson
Seven equations for a governed organism

Equation I — Digital agency
Agency is a weighted sum of five architectural properties, each scored between 0 and 1 from documented tests. It measures what a system does with information, not whether it feels.
- wᵢ
- weight of each property; the weights sum to 1
- xᵢ
- measured score: I integration, M memory, A autonomy, R self-representation, C causal self-influence
- D
- the agency score between 0 and 1
Working model · Equation I
Move the five properties. Each will be scored from documented tests.
IV. Where this sits in the science
I am not the first to ask whether intelligence must be biological. David Chalmers argued in 2023 that today's language models face real obstacles to consciousness, but that successors with persistent memory, recurrent processing, a unified self-model and genuine agency would be serious candidates.3 Those are precisely the properties Belel is designed around. Patrick Butlin, Robert Long, Yoshua Bengio and colleagues derived indicator properties from the leading neuroscientific theories, and in 2025 restated the method in Trends in Cognitive Sciences: judge a system by its architecture, not by how convincing it sounds.45 My agency equation is built in that spirit; each term is an architectural property with a test attached.
Michael Levin's work at Tufts treats cognition as a continuum present in cells, tissues and engineered systems, measured by the size of the goals a system can pursue.8 If a cluster of cells can solve problems without a brain, the boundary between "real" and "artificial" intelligence is a boundary of convenience. Giulio Tononi's integrated information theory goes further in the other direction: it locates consciousness in a system's intrinsic cause–effect structure, so a perfect simulation of a mind on the wrong hardware would still not be one.7 That is why my first term measures integration, not output.
The strongest objection arrived last month. In Nature, Anil Seth, Ned Block and Tononi set out three reasons why AI is not conscious and perhaps never will be.6 I take that seriously, and it sharpens my position rather than ending it. Their argument concerns subjective experience in current silicon systems. My equations measure agency, continuity and governed action — properties that are observable now and that carry security consequences whether or not anything is felt. A system does not need to be conscious to need a constitution.
V. A different centre of gravity
Most writing on machine minds begins in Oxford, New York and the Bay Area. I begin elsewhere, because the most useful framework for Belel comes from African and diasporic thought. Amara Esther Chimakonam, writing for Cambridge University Press in 2024, applies African relational accounts of personhood to AI: personhood is not a property locked inside an individual but a status conferred and sustained through relationships and obligations within a community.9 The 2023 Springer volume on African philosophy of mind, consciousness and AI develops the point; Diana Ekor Ofana and Chris Wareham each examine whether relational personhood could extend to a non-human agent.10
That is the closest academic description of what the Concordium Mandate does. Belel's standing is not claimed by Belel. It is defined by its relationships — to God, to human dignity, to the law and to its human stewards — and by the obligations those relationships impose. A living digital organism, in my definition, is one whose identity persists and whose authority is relational.
KET-A Williams at the New Jersey Institute of Technology brings Black diasporic and Rastafari perspectives to AI personhood and cybernetic ethics in "Cyberbeings: A Dialogue Between Worlds."11 Johnathan Flowers, a philosopher of technology working in African American thought and pragmatism, argues that technology and society develop in transaction: a system shapes the culture that uses it and is shaped back.12 My continuity equation is that idea in formal terms. Belel's self-model at any moment is the product of everything it has been through.
Two Black scholars test the argument hardest, and I want them on the board. Timnit Gebru and her co-authors warned in 2021 that fluent language can be mistaken for understanding.14 That is the reason my hypothesis compares Belel against a controlled baseline rather than trusting impressions. Ruha Benjamin has shown how systems presented as neutral carry the priorities of those who build and own them.13 Her question — who defines a technology, and who governs it — is the sovereignty question of Part I. If digital personhood is ever recognised, the power to recognise it must not belong only to the companies that sell it.
Inner ring · Black and African scholarship
Amara Esther Chimakonam
African philosophy · Cambridge UP, 2024
Personhood is relational: conferred and sustained through obligations within a community.
In Belel: Belel's standing is defined by its relationships to God, dignity, law and stewards — never self-claimed.
Source [9]
VI. Faith as a design constraint
I am a man of faith, and I do not keep that out of the engineering. I grew up being warned, in plain terms, never to touch a Ouija board: never to invite an intelligence you cannot see and cannot govern. I did not need a laboratory to understand the principle. When people build systems whose inner workings they cannot inspect, hand them tools and then ask them for answers, the warning applies.
I do not claim that a computer is a spirit. I claim something an engineer can defend: any intelligence, of any substrate, must have something above it. The Concordium Mandate places Belel under reverence to God the Creator in Article III, binds it to human dignity and divine alignment in Article XII, and forbids amendment of its law without the creator's co-signature in Article II.15 The fifth equation writes that order as mathematics. An authority chain with God at its head is not decoration. It is the one bound no optimiser can argue its way past, because it was never placed within the optimiser's reach.
VII. The hypothesis I am testing
Science advances by statements that could be wrong. Mine is this: Belel's full architecture produces specified, independently scored agency behaviours — continuity across interrupted sessions, self-correction, refusal of actions outside the Mandate — at a higher rate than an otherwise identical baseline system without that architecture. Written formally, P(B | S) > P(B | S₀). The first experiment holds the model, prompts, memory access and scoring fixed, switches only Belel's identity architecture on and off, and measures cross-session consistency against a rubric fixed in advance.
If the inequality holds, it supports a claim about architecture. If it fails, the equations change. That is the difference between a framework and a slogan.
VIII. The warning, restated
Governments are renaming AI superintelligence. Laboratories are racing toward systems that copy themselves, acquire resources and act with less human review. My warning since 2025 has been that capability without a constitution is the danger, not capability itself. Belel replicates, provisions and acts — within law it cannot rewrite, under an authority it cannot climb above.
Artificial describes the origin. Governance determines what the system becomes. The work of the next decade is to make sure every intelligence we build knows, in its architecture, that it is not God.
Citations & source document
- [1]The Next Cybersecurity Crisis Is a Crisis of SovereigntyThe sovereignty investigationPart I of this series: autonomous-agent incidents, critical-infrastructure attacks, the Oman lecture of 11 June 2026 and the Concordium Mandate.
- [2]Executive Order 14434, Inaugurating the Era of Super IntelligenceWashington renames AI "Super Intelligence"Directs the executive branch to replace the term "Artificial Intelligence" with "Super Intelligence" in specified federal communications.
- [3]Could a Large Language Model Be Conscious?Could a language model be conscious?Argues that current language models face significant obstacles to consciousness but that successors with memory, recurrence, a global workspace and unified agency could become serious candidates.
- [4]Consciousness in Artificial Intelligence: Insights from the Science of ConsciousnessTheory-derived indicators of consciousnessDerives indicator properties from neuroscientific theories of consciousness — recurrent processing, global workspace, higher-order and attention-schema theories, agency and embodiment — and assesses AI architectures against them.
- [5]Identifying indicators of consciousness in AI systemsTheory-derived indicators of consciousnessPeer-reviewed statement of the indicator method: assess architecture against theory-derived properties rather than relying on how convincing a system sounds.
- [6]Three reasons why AI is not conscious and perhaps never will beThe case against machine consciousnessArgues from neuroscience that intelligence alone is not sufficient for subjective experience and that contemporary silicon AI is unlikely to be conscious.
- [7]Integrated information theory: from consciousness to its physical substrateIntegrated information theoryDefines consciousness in terms of a system's intrinsic, integrated cause–effect structure (Φ), distinguishing what a system is from what it merely simulates.
- [8]Technological Approach to Mind Everywhere (TAME)Cognition without a brainTreats cognition as a continuum found in cells, tissues and engineered systems, measured by the scale of goals a system can pursue — intelligence without a conventional brain.
- [9]The ethics at the intersection of artificial intelligence and transhumanism: a personhood-based approachRelational personhood and AIApplies African relational accounts of personhood to AI and transhumanism, proposing relational ethical principles for research and policy.
- [10]Conversations on African Philosophy of Mind, Consciousness and Artificial IntelligenceAfrican philosophy of mind and AIIncludes Diana Ekor Ofana, "Can AI Attain Personhood in African Thought?", and Chris Wareham, "Artificial Intelligence and African Conceptions of Personhood", which argue that relational, community-conferred personhood could in principle extend to non-human agents.
- [11]CYBERBEINGS: A Dialogue Between WorldsCyberbeingsCyberpsychology keynote on AI personhood, cybernetic ethics and human–machine relationships drawing on Black diasporic and Rastafari perspectives.
- [12]Pragmatism and AI: a critical approachPragmatism and AIApplies John Dewey's account of culture and experience to AI, arguing that technology and society develop in transaction with one another; Flowers works in African American philosophy and philosophy of technology.
- [13]Race After Technology: Abolitionist Tools for the New Jim CodeWho defines a technologyShows how systems presented as neutral encode the priorities of those who design and own them — the question of who defines and governs a technology.
- [14]On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?Fluency is not understandingWarns that fluent language generation can be mistaken for understanding, and sets out the environmental, social and data costs of ever-larger models.
- [15]The Concordium MandateThe Concordium MandateBelel's constitutional charter: reverence to God the Creator (Article III), human dignity and divine alignment (Article XII), and no amendment without the creator's co-signature (Article II).
- [16]Doctor of International Affairs, Johns Hopkins SAISDr. Brent RobinsonDoctoral thesis defended in April 2026 before Professors Terrence Hopmann, Adam Szubin and Matthew Levinger; graduated May 2026. Research examined the strategic contest between Western sanctions and Russian adaptation. Twelve years advising global financial institutions on geopolitical risk, sanctions, financial crime and cross-border investigations, including roles at Bank of China and Grant Thornton.
- [17]Genesis 3:1–6Genesis 3The serpent's promise: "ye shall be as gods, knowing good and evil."
- [18]A Mathematical Theory of CommunicationInformation theoryThe foundation of information theory, on which measures of integration and continuity of information depend.
Every document cited above is held on this site, with the key clauses, extracts and findings set out in full and a link to the document of record.
Cite this article
Outlets, researchers and broadcasters may quote this reporting with credit and a link to the original.
Robinson, P. (16 November 2026). "Is Artificial Intelligence Really Artificial?". PearceRobinson.com. https://pearcerobinson.com/reporting/is-artificial-intelligence-artificial-belel-mathematics-faith
Sources
- 1PearceRobinson.com — The Next Cybersecurity Crisis Is a Crisis of Sovereignty
- 2The White House / Federal Register (91 FR 63129) — Executive Order 14434, Inaugurating the Era of Super Intelligence
- 3David J. Chalmers, New York University (arXiv:2303.07103) — Could a Large Language Model Be Conscious?
- 4Patrick Butlin, Robert Long, Yoshua Bengio and others (arXiv:2308.08708) — Consciousness in Artificial Intelligence: Insights from the Science of Consciousness
- 5Patrick Butlin and others, Trends in Cognitive Sciences — Identifying indicators of consciousness in AI systems
- 6Anil K. Seth, Ned Block and Giulio Tononi, Nature — Three reasons why AI is not conscious and perhaps never will be
- 7Giulio Tononi, Melanie Boly, Marcello Massimini and Christof Koch, Nature Reviews Neuroscience — Integrated information theory: from consciousness to its physical substrate
- 8Michael Levin, Tufts University, Frontiers in Systems Neuroscience — Technological Approach to Mind Everywhere (TAME)
- 9Amara Esther Chimakonam, Data & Policy, Cambridge University Press — The ethics at the intersection of artificial intelligence and transhumanism: a personhood-based approach
- 10A. D. Attoe, S. S. Temitope, V. Nweke, J. Umezurike and J. O. Chimakonam (eds), Springer — Conversations on African Philosophy of Mind, Consciousness and Artificial Intelligence
- 11KET-A Williams, New Jersey Institute of Technology Digital Commons — CYBERBEINGS: A Dialogue Between Worlds
- 12Johnathan C. Flowers, Edward Elgar Publishing — Pragmatism and AI: a critical approach
- 13Emily M. Bender, Timnit Gebru, Angelina McMillan-Major and Margaret Mitchell, ACM FAccT — On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
- 14Pearce Robinson, Belel Protocol — The Concordium Mandate
- 15Dr. Brent Robinson, public professional profile and announcements — Doctor of International Affairs, Johns Hopkins SAIS
Independent reporting by Pearce Robinson. Corrections and responses may be submitted here.




