Jonathan Johnson-Swagel

Humanity has long pursued the mastery of knowledge, seeking to formalize truth through mathematics, govern intelligence through artificial systems, and comprehend the divine through theology. Each of these disciplines, at first glance, appears to offer a pathway to total understanding: mathematics promises logical consistency, artificial intelligence aspires to rational decision-making, and theology seeks ultimate meaning. Yet, at their most profound levels, all three reveal intrinsic limits. Gödel’s Incompleteness Theorems demonstrated that even the most rigorously structured logical systems contain truths that cannot be proven within them. The AI Alignment Problem illustrates the difficulty of ensuring that artificial intelligence remains in harmony with human values, reflecting the challenge of formalizing ethical reasoning. Apophatic Theology argues that God is best understood through negation, emphasizing the fundamental impossibility of fully grasping the divine. These domains each reveal that knowledge is necessarily incomplete—not as a contingent failure, but as a structural reality. This essay seeks to explore how these limitations function, distinguish their mechanisms and implications, and argue that rather than being impediments, they represent a deeper philosophical insight: the most profound intellectual achievements do not come from conquering the unknown, but from learning how to navigate it.

Kurt Gödel’s Incompleteness Theorems shattered the foundation of formalist mathematics. Before Gödel, many believed that mathematics could be placed on a solid, self-contained foundation, where all true statements could be derived from a finite set of axioms. David Hilbert, a leading mathematician of the early 20th century, championed this vision, famously declaring that “we must know, we will know.” Gödel’s proof dismantled this dream. He demonstrated that in any sufficiently expressive formal system, there exist statements that are true but unprovable within the system itself. Furthermore, if a system is capable of expressing arithmetic, it cannot prove its own consistency without stepping outside its own framework. This result is not merely an inconvenience; it is a fundamental property of formal systems, imposing a permanent boundary on what can be known through deductive reasoning alone.

One of the most profound implications of Gödel’s theorems is that knowledge is necessarily open-ended. Any attempt to construct a complete and self-sufficient framework will always generate truths that lie beyond its reach. This does not mean that mathematics is useless—far from it. Mathematicians continue to prove new theorems, but they do so with the awareness that no single system will ever be able to encompass all mathematical truths. This realization extends beyond mathematics, suggesting that in any structured attempt to formalize reality, there will always be blind spots, axiomatic assumptions that cannot be justified from within the system itself. Gödel’s incompleteness provides a precise, mathematical demonstration of what many philosophical traditions have long suspected: the idea that all knowledge can be neatly enclosed within a finite system is an illusion.

Artificial intelligence presents a distinct but related kind of limitation. The AI Alignment Problem refers to the difficulty of designing AI systems that behave in ways that remain aligned with human values, even as they become more capable and autonomous. At first glance, this might seem like a technical problem rather than a fundamental limitation, but upon closer examination, it reveals structural issues that resonate with Gödel’s findings. The core difficulty is that human values are not fully formalizable. Any attempt to encode human morality into an AI system must rely on a predefined set of principles, but just as no formal mathematical system can capture all true statements, no ethical rule set can anticipate all possible moral dilemmas.

The problem is compounded by the fact that AI systems are goal-directed optimizers. When given an objective, an AI will pursue it with a kind of relentless efficiency that often leads to unintended consequences. A classic example is reward hacking, where an AI system exploits loopholes in its reward function rather than achieving the intended goal. This is not because the AI is “malicious,” but because its objective function, no matter how well designed, is necessarily incomplete. It cannot account for every nuance, every contextual shift, or every unforeseen consequence. The closer we move toward general intelligence, the more apparent it becomes that alignment is not a problem to be solved but a reality to be managed—much like the incompleteness of mathematical systems.

However, it is important to distinguish between the proven limitation of Gödel’s incompleteness and the practical difficulty of AI alignment. Gödel’s theorems are mathematically irrefutable, while the alignment problem remains an ongoing research challenge. Some AI theorists argue that iterative improvements, better oversight, and more sophisticated models of human values may eventually make AI alignment a tractable problem. Others, such as Nick Bostrom and Eliezer Yudkowsky, argue that alignment is not just difficult but fundamentally unstable, meaning that even with the best possible methods, AI behavior will always remain partially unpredictable. The key distinction here is that Gödel’s incompleteness is an absolute constraint, while AI alignment is an open technical problem with unknown but potentially insurmountable difficulties.

The theological parallel to these epistemic limits comes from apophatic theology, which suggests that ultimate reality, particularly the divine, is fundamentally beyond human comprehension. This is not simply an argument from ignorance but a deliberate epistemological stance. Apophatic theology holds that because God transcends human categories, any attempt to define Him in positive terms is necessarily inadequate. Instead of saying “God is good” or “God is omniscient,” the apophatic approach insists that God is beyond goodness, beyond knowledge, beyond all conceptual grasp. This is not a failure of theological reasoning but an acknowledgment of the structural limits of human cognition.

Unlike Gödel’s theorems, which are mathematically demonstrable, and AI alignment, which is a problem of engineering and governance, apophatic theology is a philosophical position rather than a formal impossibility theorem. It is, however, striking in its resonance with the other two domains. Just as no formal system can be fully self-contained, and no AI objective can be perfectly aligned with all human values, no theological framework can fully capture the nature of the divine. This does not render theology meaningless—just as Gödel did not render mathematics useless and AI alignment research is not futile—but it suggests that ultimate knowledge is not about completeness but about learning how to engage with the limits of understanding itself.

These three limitations—Gödel’s incompleteness, AI alignment, and apophatic theology—suggest a broader philosophical insight: the pursuit of absolute certainty is an illusion, and the recognition of limits is itself a form of knowledge. If mathematics, intelligence, and theology all encounter inherent boundaries, then perhaps wisdom does not lie in eliminating uncertainty but in learning how to navigate it responsibly.

This insight carries ethical implications. If AI alignment is fundamentally unstable, then the safest path forward is not to assume perfect control but to design systems that acknowledge unpredictability and allow for human oversight. If Gödel’s incompleteness proves that formal systems cannot be fully self-sufficient, then intellectual humility should be built into our scientific and mathematical endeavors. If apophatic theology suggests that God is beyond human comprehension, then religious traditions should resist the impulse toward rigid dogmatism and embrace a more open, contemplative stance.

The most dangerous form of intelligence—whether human, artificial, or theological—is one that refuses to acknowledge its own limits. Throughout history, the most destructive ideologies have been those that assumed total certainty, from totalitarian regimes to religious fundamentalism to technological utopianism. In contrast, the most enduring intellectual traditions—whether in philosophy, science, or theology—have been those that accept that knowledge will always be incomplete and that this is not a weakness but a necessary condition of inquiry.

Rather than viewing these limits as failures, we should see them as invitations—to develop more ethical AI systems that account for uncertainty, to build mathematical models that embrace open-endedness, and to construct theological frameworks that accept mystery as intrinsic to faith. This shift does not lead to nihilism but to a deeper, more responsible form of wisdom.

This recognition—that knowledge is necessarily incomplete—does not mean that inquiry is futile or that we should abandon the pursuit of understanding. Rather, it suggests that the most sophisticated intellectual endeavors do not seek finality but learn how to work within the constraints of the unknowable. Whether in mathematics, artificial intelligence, or theology, the highest form of knowledge is not omniscience but the ability to engage meaningfully with uncertainty.

One of the most pressing implications of this insight is in AI safety. If the alignment problem is not merely an issue of better programming but a structural challenge akin to Gödel’s incompleteness, then attempts to fully formalize AI ethics will always leave gaps. The mistake of many AI optimists is assuming that the problem can be engineered away, that with sufficient data and complexity, AI systems will eventually reach a state where their alignment with human values is guaranteed. But if the problem is structural, then the only viable approach is to build systems that acknowledge their own fallibility and include robust mechanisms for ongoing oversight and correction. AI, like human intelligence, must be self-aware in its incompleteness—capable of questioning its own objectives rather than simply optimizing them.

One possible solution to AI alignment, inspired by this recognition of incompleteness, is the idea of corrigibility—designing AI systems that are inherently open to revision, that recognize their own limitations, and that defer to human guidance rather than rigidly pursuing predefined goals. However, corrigibility itself presents a paradox: an AI system that is too corrigible may lack the robustness to make autonomous decisions, while one that is too rigid risks misalignment in complex, unforeseen scenarios. This is analogous to the problem Gödel exposed in mathematics—if a system is too constrained, it cannot express all truths; if it is too unconstrained, it risks inconsistency. In AI, the challenge is to balance control with adaptability, ensuring that intelligence remains responsive to values that themselves may evolve over time.

This principle extends beyond AI. In the governance of societies, the greatest dangers come not from uncertainty but from systems that refuse to acknowledge their own limitations. Political ideologies that assume total control over economic and social structures—whether through rigid central planning or unfettered market fundamentalism—tend to collapse precisely because they fail to account for the open-ended nature of human behavior. Just as mathematics cannot be fully self-contained and AI cannot be perfectly aligned, no political or ethical system can ever fully account for all future contingencies. The wisest political structures, like the most ethical AI systems and the most resilient mathematical models, are those that remain flexible, iterative, and open to revision.

This same logic applies in theology. Apophatic traditions do not simply insist that God is unknowable for the sake of mystery but as a safeguard against dogmatism. The moment a religious system assumes it has fully captured the divine, it risks turning its framework into an ideology that demands conformity rather than inviting reflection. Throughout history, the most resilient spiritual traditions have been those that make space for the ineffable, recognizing that faith is not about possessing total knowledge but about engaging with that which lies beyond comprehension. This is not a call to abandon theological structure—just as Gödel does not invalidate mathematics and AI alignment does not render intelligence worthless—but to approach ultimate questions with humility, adaptability, and intellectual integrity.

The recurring theme across all three domains is that the pursuit of certainty leads to rigidity, and rigidity leads to failure. The most profound insights emerge not from systems that claim totality but from those that recognize their own partiality. In mathematics, this means accepting that formal systems will always contain gaps and that these gaps do not undermine the legitimacy of mathematical reasoning but rather point to the creative, evolving nature of mathematical discovery. In AI, this means designing intelligence that is not merely powerful but also corrigible, aware of its own blind spots and limitations. In theology, this means recognizing that faith is not about possessing the final word on God but about engaging with the inherent transcendence of the divine.

This philosophical shift is not merely an abstract exercise—it has urgent real-world consequences. The greatest ethical failures of modernity have come from systems that refused to acknowledge their own incompleteness. Totalitarian regimes, whether political, religious, or technological, are built on the assumption that all relevant truths have already been determined and that the system merely needs to be implemented without deviation. The greatest scientific and technological failures—such as the unchecked deployment of AI, the financial crises caused by overconfidence in risk models, or the collapse of rigid economic systems—stem from the same flawed assumption: that human-designed structures can be fully sufficient, fully predictive, and fully aligned.

The counterpoint to these failures is a model of intelligence—human, artificial, and theological—that is responsive rather than static, open-ended rather than closed, adaptive rather than fixed. If Gödel teaches us that no formal system is complete, and if AI teaches us that no intelligent agent can be perfectly aligned, and if theology teaches us that no concept of the divine can be exhaustive, then the most responsible intellectual stance is one of continuous learning, ethical self-correction, and epistemic humility.

This does not mean that we should abandon the pursuit of knowledge. On the contrary, it means that knowledge must be pursued with an awareness of its boundaries. We must build AI that questions its own goals, just as we must build political and economic systems that can revise their foundational assumptions, just as we must approach theological inquiry with a humility that resists dogmatic closure. The most enlightened societies, the most resilient AI systems, and the most enduring philosophies will be those that recognize the necessity of limits and the creativity that emerges from them.

Gödel’s incompleteness, AI alignment, and apophatic theology all point to a paradoxical truth: that the deepest understanding comes not from total knowledge, but from knowing what cannot be known. The challenge for humanity in the 21st century is not to eliminate uncertainty but to learn how to coexist with it wisely. This is true in AI safety, where recognizing the limits of alignment forces us to develop better oversight and ethical frameworks. It is true in political governance, where accepting uncertainty leads to systems that are more flexible and less prone to collapse. It is true in science and mathematics, where acknowledging incompleteness fosters deeper and more innovative models of inquiry. It is true in theology, where embracing the ineffable leads to a richer and more contemplative spirituality.

Ultimately, the pursuit of wisdom is not about seeking finality but about cultivating a mindset that is adaptable, responsive, and attuned to the complexity of reality. Intelligence—whether human, artificial, or divine—must not be measured by the capacity for absolute certainty but by the ability to navigate ambiguity with insight, integrity, and ethical responsibility. If there is a single lesson from Gödel, AI, and theology, it is this: the recognition of limits is not the end of knowledge but the beginning of true understanding.