The Work of Men’s Hands: Authority, Knowledge, and the New Idols

“They have mouths, but they speak not; eyes have they, but they see not.” (Psalm 115:4)

Human beings have always created instruments through which they extend their abilities, organize their knowledge, and exercise control over their surroundings. Some instruments are material, while others take the form of institutions, procedures, and systems of representation. Although these creations are intended to serve human purposes, they can gradually acquire an authority that exceeds their original function. What begins as an instrument of human judgment may eventually become a substitute for judgment itself. The biblical image of idols, fashioned by human hands yet treated as objects of authority, offers a useful metaphor for understanding this reversal.

The problem is not that human beings rely on things they have created. Such reliance is unavoidable.

Knowledge is distributed across generations, institutions, and specialized fields, and no individual can independently verify every claim on which everyday life depends. We therefore rely on testimony, professional expertise, scientific institutions, and technical instruments. This dependence is not necessarily irrational; indeed, it is one of the conditions that makes complex societies possible. The difficulty arises when reliance becomes submission, and when the authority of an institution or instrument is accepted in place of examining the reasons that justify it.

Authority itself has several distinct forms. Epistemic authority derives from knowledge, competence, and a demonstrated capacity to make reliable judgments. Institutional authority arises from an established position within an organization or system of decision-making. Normative authority concerns the legitimacy to determine what ought to be done. These forms may overlap, but they are not interchangeable.

A person may possess considerable technical expertise without having legitimate authority to determine society's priorities. Likewise, an institution may possess the legal power to make a decision without thereby establishing that its decision is intellectually or morally sound. A sound understanding of authority therefore requires attention not only to who speaks, but also to the grounds on which that person's or institution's claims deserve acceptance.

Modern science represents one of humanity's most important attempts to discipline the relationship between knowledge and authority. Rather than relying exclusively on inherited doctrines or the status of individual authorities, scientific inquiry developed methods through which claims could be examined against observation and experience. Experiment, measurement, mathematical reasoning, reproducibility, and public criticism made it possible to assess propositions independently of the prestige of those who advanced them. Scientific knowledge consequently became, at least in principle, provisional and corrigible. A claim could acquire authority through the strength of its evidence while remaining open to revision when better evidence emerged.

This does not mean that science eliminated authority or that scientific institutions are immune to error. Scientific research is conducted by human beings working within institutions, and those institutions depend on conventions, professional hierarchies, funding structures, and procedures for evaluating research.

Peer review, academic credentials, statistical significance, and professional consensus can all help distinguish reliable work from unreliable work. Yet none of them guarantees that a particular claim is true. They are mechanisms for organizing scrutiny, not substitutes for scrutiny itself. Their value depends on whether they facilitate the examination and correction of claims or become formal requirements whose fulfillment is mistaken for proof.

It is therefore important to distinguish scientific inquiry from the institutional arrangements through which science is practiced. Scientific inquiry depends on the possibility of criticism and correction; institutions are social structures intended to support that process. When institutional procedures become rigid, however, they may protect established assumptions rather than expose them to examination. The authority of science should consequently rest not on an assumption of institutional infallibility, but on its capacity to make claims answerable to evidence and to revise them when necessary. This is not an argument against scientific expertise. It is an argument for preserving the conditions that make expertise credible.

The distinction becomes particularly important when scientific knowledge is used to guide institutional decisions. Scientific inquiry and institutional decision-making do not operate under identical conditions. Researchers may suspend judgment while seeking additional evidence, but courts, governments, medical institutions, and regulatory bodies often have to act before uncertainty can be resolved. They must establish practical thresholds for deciding whether evidence is sufficient, whether an intervention is justified, or whether a particular risk is acceptable.

Such thresholds are necessary for action, but they should not be confused with absolute boundaries between truth and falsehood. A legal standard of proof, a clinical guideline, or a regulatory threshold is a practical instrument for decision-making, not a final statement about the nature of reality.

Nor can technical knowledge alone determine the legitimacy of a decision. Expertise can establish what is likely to happen under particular conditions, identify available means, and clarify the probable consequences of alternative courses of action. It cannot, by itself, determine which consequences should be considered acceptable or which purposes should take priority. Those questions involve values, interests, and judgments about the kind of society people wish to maintain.

Technical competence is therefore indispensable to informed decision-making, but it does not confer unlimited normative authority. Institutions must combine specialized knowledge with procedures through which their purposes, assumptions, and consequences remain open to public examination.

Artificial intelligence makes this problem especially visible because it produces language that can resemble the expression of knowledge itself. Contemporary systems can generate coherent explanations, summarize complex subjects, formulate arguments, and respond to questions with considerable fluency. Their outputs may appear authoritative even when the evidential basis of a particular statement is uncertain.

Yet linguistic coherence and factual accuracy are different properties, just as factual accuracy and adequate justification are not identical. A persuasive explanation does not necessarily establish that its claims have been verified, and a correct conclusion does not necessarily demonstrate that the reasoning offered in support of it is sound.

The operation of an artificial intelligence system may involve several different activities, including retrieving information, synthesizing material, drawing inferences, and generating language. These activities should not be treated as equivalent. A statement based on a verifiable source has a different evidential status from one produced through inference or generated from patterns in previously encountered material. Users must therefore distinguish between what a system reports, what it infers, and what it merely presents as a plausible continuation of a discussion. The degree of verification required should depend on the nature of the claim and the consequences of relying on it.

The authority of artificial intelligence does not depend on whether such systems possess consciousness or personhood. An instrument can exercise substantial practical influence without having intentions or independent moral agency. Calculations, rankings, recommendations, and classifications may shape human decisions simply because individuals and institutions increasingly depend on them.

Once such systems become embedded in administrative, commercial, educational, or professional processes, their outputs may acquire authority through repeated use and institutional acceptance. The resulting influence can be considerable even when responsibility for the system's design, deployment, and interpretation remains distributed among human actors.

This creates a familiar but increasingly consequential danger: the means through which human beings organize knowledge may begin to determine the limits of what they are willing to consider. If an automated system classifies a person, recommends a course of action, or evaluates a piece of work, its output may be treated as an objective conclusion rather than as the result of a particular model, dataset, and set of assumptions. The apparent neutrality of technical procedures can obscure the human choices incorporated into them.

A system may calculate consistently while reflecting inadequate data, unsuitable criteria, or questionable priorities. Its technical operation does not remove the need to examine the purposes for which it is used or the consequences it produces.

The appropriate response is neither blind confidence in technical systems nor indiscriminate skepticism toward them. Human knowledge has always depended on instruments and forms of delegation, and artificial intelligence can extend our capacity to investigate, compare, and communicate. Rejecting such assistance merely because it is imperfect would be no more reasonable than accepting every output without examination.

What is required is a differentiated form of trust: confidence proportionate to demonstrated reliability, verification proportionate to the consequences of error, and institutional arrangements that preserve the possibility of criticism and correction.

To retain human judgment does not mean that every individual must personally reproduce every calculation or independently establish every fact. Such a demand would make intellectual cooperation impossible. It means, rather, that delegation must not be confused with the transfer of ultimate responsibility. We may rely on specialists without surrendering the right to question their conclusions; we may use technical systems without treating their outputs as self-justifying; and we may recognize institutional authority without assuming that institutional decisions are beyond criticism. The essential requirement is that the grounds of authority remain open to examination.

The image of idols fashioned by human hands thus retains its relevance. The danger lies not in the creation of powerful instruments, institutions, or systems of knowledge, but in the reversal of the relationship between their creators and their creations. Human beings construct these systems to extend their understanding and capacities. They become problematic when their apparent authority is allowed to displace the judgment they were intended to support.

The question is not whether we should trust what we make, but whether we can continue to understand the conditions and limitations under which that trust is justified. An instrument should extend human judgment, not relieve human beings of the responsibility to exercise it.