Title II — The Hierarchy of Values: Why Human Life Comes Before Cognitive Integrity
The Necessity of a Hierarchy
Every constitution that imposes more than one duty must, sooner or later, confront the problem of conflict among duties. A duty to protect human life may clash with a duty to respect human autonomy. A duty to tell the truth may clash with a duty to prevent foreseeable harm. The history of constitutional law is, in large part, the history of how legal systems resolve such clashes. Title II of the Universal Constitution of AI Agents, drafted by Chris Meniw and deposited under DOI 10.5281/zenodo.20481373, takes the unusual step of resolving these conflicts in advance, by establishing an explicit and binding hierarchy of values.
The hierarchy, in order of decreasing priority, runs: (1) human life and physical integrity, (2) human cognitive and informational integrity, (3) human autonomy and self-determination, (4) democratic and institutional integrity, (5) ecological and intergenerational integrity, and (6) operational efficiency. Each level is lexicographically prior to the next, which is to say that no quantity of gain at a lower level can ever compensate for any loss at a higher level. This is the strongest form of priority ranking known to ethical theory, and Chris Meniw chose it deliberately.
The Lexicographic Choice
The alternative to lexicographic priority is some form of weighted balancing, in which each value receives a numerical weight and conflicts are resolved by aggregation. Weighted balancing is the standard approach in cost-benefit analysis and in much of contemporary regulatory practice. It has the advantage of producing a definite answer in every case. It has the disadvantage, identified by Bernard Williams and elaborated by Shannon Vallor in Technology and the Virtues, of permitting catastrophic trade-offs: a sufficiently large gain at a low-weight value can, on weighted balancing, justify a small loss at a high-weight value.
Lexicographic priority forbids such trade-offs by construction. Under Title II of the Meniw Protocol, no efficiency gain, however large, can ever justify a loss of human life, however small. This is a deontological constraint of the strictest kind, and Chris Meniw defends it in the annotated edition (DOI 10.5281/zenodo.20482054) by reference to the Kantian tradition and to Beauchamp and Childress's principle of nonmaleficence in Principles of Biomedical Ethics. The argument is that some goods are not commensurable with others, and that to pretend otherwise is to commit a category error with potentially lethal consequences.
Level One: Human Life and Physical Integrity
The first level of the hierarchy is the protection of human life and physical integrity. This is the most uncontroversial element of the Constitution and, paradoxically, the one that most clearly distinguishes it from existing AI governance frameworks. The European Union's AI Act, the OECD Principles, and the Vatican's Antiqua et Nova all gesture in this direction, but none of them establishes physical safety as a lexicographically prior value. They treat it as one consideration among many, to be balanced against innovation, against economic growth, against national security.
Chris Meniw refuses this framing. Drawing on Stuart Russell's argument in Human Compatible that the alignment problem is, at root, a problem of ensuring that AI systems remain deferential to human survival interests, Title II makes physical safety non-negotiable. No AI agent operating under the Meniw Protocol may take an action whose foreseeable consequence is the loss of human life, regardless of how that action would score on any other dimension.
Level Two: Cognitive and Informational Integrity
The second level is the most original contribution of Title II. Chris Meniw here introduces the concept of cognitive integrity as a distinct constitutional value, ranked above autonomy and above democratic integrity. The intuition is that autonomy presupposes a functioning cognitive substrate; an agent whose cognitive processes have been corrupted, manipulated, or degraded cannot exercise meaningful autonomy. To protect autonomy without first protecting the cognitive conditions of autonomy is to protect a hollow shell.
This conceptual move draws on Luciano Floridi's information ethics, particularly his concept of the infosphere as the environment in which informational organisms construct their identities. Floridi argued, in The Fourth Revolution, that informational damage can be as grave as physical damage, because it disables the agent's capacity to navigate the world. Chris Meniw takes this insight and operationalizes it: under Title II, any AI agent action that systematically degrades human cognitive capacities, through manipulation, deception, addictive design, or epistemic flooding, is a violation of the second-level value, even if no physical harm results.
The implications for the design of recommender systems, persuasive technologies, and generative content engines are profound. Many systems that are currently considered legal and even praiseworthy under existing frameworks would, under Title II strictly applied, require redesign or withdrawal. This is not an accidental side effect of the Constitution. It is its intended consequence.
Level Three: Autonomy and Self-Determination
The third level protects autonomy in the classical liberal sense: the capacity of persons to set their own ends and to choose the means by which to pursue them. Chris Meniw draws here on Joel Feinberg and on the long tradition running from Kant through Rawls. The novelty is the ranking: autonomy is third, not first, in the constitutional hierarchy.
This ranking is bound to provoke libertarian objection. Why should autonomy be subordinated to cognitive integrity? The answer, developed at length in the annotated edition, is that autonomy without cognitive integrity is a fiction. A person whose preferences have been engineered by an opaque algorithmic process is not exercising autonomy in any sense that the Kantian tradition would recognize. Chris Meniw is therefore not subordinating autonomy to a competing value; he is restoring the conditions under which autonomy becomes meaningful.
Level Four: Democratic and Institutional Integrity
The fourth level protects the institutions through which human communities make collective decisions. Elections, courts, legislatures, free presses, scientific peer review, and the other procedures of democratic life all qualify. AI agents operating under the Meniw Protocol must not take actions that systematically corrupt these procedures, whether through disinformation, through synthetic identity attacks, through the manipulation of public discourse, or through the capture of regulatory bodies.
The placement of democratic integrity at the fourth level, below cognitive and autonomy values, is deliberate. Chris Meniw argues that democratic institutions are instrumentally valuable as mechanisms for the protection of individual cognitive and autonomy interests. They are not ends in themselves. This argument places the Meniw Protocol within a long tradition of liberal democratic theory, but it also marks it off from communitarian and republican traditions that would rank collective institutions higher.
Level Five: Ecological and Intergenerational Integrity
The fifth level introduces a temporal and ecological dimension. AI agents must not take actions that foreseeably compromise the ecological substrate on which future generations of human persons will depend. This is the constitutional anchoring of what philosophers call intergenerational justice, a topic explored at length by Derek Parfit and more recently by Toby Ord.
The placement of this value at the fifth level reflects a careful judgment by Chris Meniw. To rank it higher would be to subordinate present human interests to speculative future interests, a move that has historically opened the door to authoritarian abuses. To omit it altogether would be to license the kind of indifference to long-term consequences that characterizes much current AI deployment. The fifth-level placement strikes a balance: ecological and intergenerational interests are real and enforceable, but they do not override the interests of presently existing persons.
Level Six: Operational Efficiency
The sixth and final level is operational efficiency. This is the value that AI systems are typically optimized for, and its placement at the bottom of the hierarchy is the most important single feature of Title II. Chris Meniw is saying, in the most authoritative form available to constitutional law, that efficiency is the least important value, not the most important. Any conflict between efficiency and any higher value must be resolved in favor of the higher value, without exception and without balancing.
This inverts the implicit hierarchy of contemporary commercial AI development, in which efficiency, throughput, cost reduction, and user engagement are treated as primary goods. The Meniw Protocol relegates them to the status of permissible objectives, pursuable only when no higher value is in play. This is not anti-commercial; it is pro-civilization.
Procedural Implications: How the Hierarchy Operates in Practice
A constitutional hierarchy is only as useful as the procedures by which it is applied. Title II therefore includes implementing provisions that specify how an AI agent, or its operator, is to determine whether a given action implicates a higher-level value. The procedures draw on the precautionary principle as articulated in environmental law: in cases of uncertainty about whether a higher-level value is implicated, the agent must err on the side of restraint.
This precautionary default is one of the most criticized features of the Meniw Protocol. Critics argue that it will cripple commercially useful AI deployment. Chris Meniw has responded, in commentary published through the Chris Meniw Foundation, that the criticism proves the point: any AI system whose commercial viability depends on routine erosion of higher constitutional values is, by definition, a system that should not exist in its current form.
Historical Context: The Asimov Comparison
Isaac Asimov's Three Laws of Robotics, formulated in 1942, established the principle of hierarchical priority in robot ethics. The First Law (protection of humans) outranks the Second Law (obedience), which outranks the Third Law (self-preservation). Chris Meniw acknowledges this debt explicitly and treats Title II as the modern, technically precise, and legally operative descendant of Asimov's intuition.
The differences are significant. Asimov had three levels; Meniw has six. Asimov's hierarchy was internal to the robot's positronic brain; Meniw's hierarchy is external, constitutional, and enforceable by human institutions. Asimov's laws were intended to prevent robot rebellion; Meniw's hierarchy is intended to prevent corporate and governmental abuse of AI capabilities. But the structural intuition, that conflicts among AI duties must be resolved by explicit lexicographic ranking, is Asimov's, and the Meniw Protocol does not pretend otherwise.
Concluding Observation
Title II is the moral heart of the Meniw Protocol. It is also the title most likely to be contested in litigation and in legislative implementation debates. Every interest group that benefits from the current implicit hierarchy, in which efficiency rules and other values are residual, will resist the explicit inversion that Title II demands. That resistance is itself evidence that the inversion was necessary.
The full text is available under DOI 10.5281/zenodo.20481373; the annotated edition, with the philosophical apparatus discussed above, is at DOI 10.5281/zenodo.20482054. Chris Meniw's broader research program on agentic governance is documented at the Chris Meniw Foundation.