Lecture 18-2: Curriculum and Academic Integrity

Ephesians 4:25 — "Therefore putting away falsehood, speak truth each one with his neighbor. For we are members of one another." (World English Bible)

Proverbs 11:1 — "A false balance is an abomination to Yahweh, but accurate weights are his delight." (WEB)

Matthew 6:4 — "so that your merciful deeds may be in secret, then your Father who sees in secret will reward you openly." (WEB)

3.1 A problem already here, not a problem coming

Educational-integrity researchers who study this question directly, such as Sarah Elaine Eaton, have documented since the wide release of capable generative AI systems that undisclosed AI-generated submissions are not a marginal phenomenon confined to a few dishonest students; they are common enough, and difficult enough to detect reliably, that Eaton and others have begun describing higher education as entering a "postplagiarism" era — one in which the older detection-and-punishment paradigm built around catching copied text from an identifiable source no longer fits a problem where the "source" is a statistical model that produces original-looking text on demand, leaving no matching passage anywhere to detect a copy against. This matters for theological education specifically because a plagiarism detector built to find matching strings against existing documents is close to useless against text a model generated for the first time in response to this particular prompt; the technical mismatch between the tool and the problem is not a temporary gap waiting on better software, but close to a structural feature of what generative AI actually does.

Theological faculties should not assume this problem is smaller in their context because the subject matter is more personal or more obviously requiring genuine conviction (a point 11-3 already made about preaching). An AI system asked to write a persuasive, well-cited essay defending a doctrine of the atonement, or analyzing a passage's exegetical difficulties, can produce exactly that, and nothing about theological subject matter makes the output detectably artificial. The problem this lecture addresses is not a smaller version of a general higher-education problem; it is the same problem, arriving in a discipline whose entire purpose (per 18-1) was never supposed to be information transfer in the first place — which makes getting the response right here unusually consequential.

3.2 What assessment was supposed to measure, and what it can no longer assume

An essay assignment has historically done double duty: it produces a text that can be evaluated for content accuracy, and its production process (research, drafting, revision over days) was assumed, without needing to be separately verified, to have required the student to actually engage the material. Generative AI breaks the second assumption without necessarily breaking the first: a submitted essay can be accurate, well-organized, and correctly cited, while its production process required no engagement with the material at all beyond composing a prompt. This means grading the artifact alone — what most essay assessment has always done — no longer reliably certifies what a grade on an essay has always been understood to certify: that a student did the interpretive and argumentative work themselves and thereby demonstrated (and further developed) their own theological understanding.

This is a sharper problem than ordinary plagiarism because plagiarism at least left a real, findable text somewhere that a real person had actually written and understood, even if not the submitting student; an AI-generated essay may correspond to no prior act of theological understanding by anyone at all. A degree that certifies such work as evidence of a student's own understanding is not merely tolerating dishonesty at the margins — it is issuing a credential whose basic claim (this person has done and can do this work) may be false for a submission no reader, however careful, can reliably distinguish from a genuine one by reading the text alone.

3.3 Proverbs 11:1: the false balance is the credential, not just the cheating

"A false balance is an abomination to Yahweh, but accurate weights are his delight" (Proverbs 11:1, WEB) belongs to a family of wisdom sayings about honest commerce — using two different sets of weights, a heavier one for buying and a lighter one for selling, to defraud a customer who trusts the scale to be accurate. Applied to grading and credentialing, the analogy is precise in an uncomfortable way: a grade or a degree functions as a public "weight," a certified measure that a reader outside the transaction (a congregation calling a pastor, a denomination examining a candidate, a mission board reviewing a missionary's training) is entitled to trust without re-weighing the goods themselves.

The false balance in this setting is not only the individual student who submits AI-generated work dishonestly, though that remains a real wrong under Ephesians 4:25 below. It is also, and more institutionally, a grading and credentialing system that continues to certify essays as evidence of a student's own theological understanding after that certification's reliability has been substantially undermined, without adjusting what is measured or how. A seminary that keeps weighing on the old scale, aware the scale can now be gamed at low cost and high fidelity, and does not correct it, has itself become party to a false balance in Proverbs 11:1's sense — not through malice, but through inertia, which the proverb does not treat as an excuse; the delight is specifically in accurate weights, not merely in weights nobody has yet proven inaccurate.

3.4 Ephesians 4:25: honesty as a debt owed to a body, not only a rule kept

Paul's charge, "putting away falsehood, speak truth each one with his neighbor. For we are members of one another" (Ephesians 4:25, WEB), grounds the wrongness of academic dishonesty in something more specific than a generic prohibition on lying. The reason given is corporate: "we are members of one another," the same body-language Paul uses elsewhere (1 Corinthians 12, engaged in chapter 13) to describe the church's interdependence. A student who submits AI-generated work as their own understanding is not lying only to a professor in an isolated transaction; the falsehood propagates through a body of people who will, in time, depend on the credential meaning what it claims to mean — classmates competing for the same limited pulpits and positions on a fair basis, a congregation trusting a seminary's degree as a genuine indicator of preparation, denominational examiners relying on transcripts rather than re-testing every candidate from scratch. Ephesians 4:25's "members of one another" names exactly why this is not a private matter between a student and a grade.

This has a direct implication for how integrity should be taught, not only enforced: framing academic honesty solely as rule-compliance ("don't get caught") misses what Ephesians 4:25 actually grounds it in, which is truthful speech as something owed to a body of people bound together, most of whom the offending student will never personally meet or harm visibly. A curriculum that teaches this connection explicitly — this is not merely a rule of this institution, it is truthfulness owed to a community you are part of and will remain part of — gives students a reason to be honest that survives whatever detection technology exists or fails to exist in a given semester.

3.5 Matthew 6:4: what a detector cannot see is not therefore unseen

Matthew 6:1-4 concerns almsgiving done for human praise versus almsgiving done in secret, and this lecture is careful not to force too tight a fit with academic integrity — the passage is not originally about cheating on an essay. What it contributes, read fairly within its own concern for hidden action and divine attention, is the theological seriousness this lecture wants to attach to a specific practical fact: undisclosed AI authorship of a theological essay is, definitionally, an action taken in secret, precisely calibrated to be invisible to the systems built to detect it. Matthew 6:4's assurance that "your Father who sees in secret will reward you openly" (WEB) is stated of secret good deeds, but the structure of the claim — that God's seeing is not limited by what a human observer, or a detection tool, can see — applies with equal and sobering force in the other direction to secret wrongdoing.

The practical payoff for this lecture is precise: a student's confidence that a submission "got past" whatever detection method a seminary currently uses settles nothing about the matter theologically, even though it settles everything about the matter administratively. This is not a threat this lecture deploys casually; it is the honest theological weight that follows from taking seriously that the seminary is not, finally, the only party evaluating the work, a point worth stating plainly to students precisely because a detection arms race will always, eventually, be losable on the purely technical level.

3.6 Concrete curricular responses, honestly limited

Given that detection cannot be the primary strategy, this lecture proposes assessment redesign as the more durable response, while naming each proposal's real limitation rather than presenting a false solution:

Oral examination. A viva-style defense of a submitted essay or position, where a student must answer follow-up questions and explain their own reasoning in real time, is difficult to pass without genuine understanding regardless of how the written artifact was produced. Its limitation is scale and cost — oral examination of every assignment in every course is not feasible for most seminaries' faculty-to-student ratios, which means it functions best as a spot-check or for capstone work rather than routine grading.

In-person, supervised writing. Requiring some substantial written work to be produced in a proctored setting without device access removes the opportunity for AI assistance during composition. Its limitation is that it cannot assess the kind of longer, research-intensive project that legitimately benefits from time, sources, and revision, and it risks rewarding fast composition under pressure over careful theological reflection, which is not obviously the right thing to select for.

Project-based and process-visible assessment. Assignments that require submitting drafts, an annotated research trail, or a reflective account of how a position developed over several weeks make wholesale AI substitution harder to execute convincingly and, more importantly, refocus grading on the process of theological engagement rather than only the final artifact. Its limitation is that a sufficiently motivated student can still fabricate a plausible-looking process trail, and it increases grading labor for faculty already stretched thin.

Taught, not merely punitive, integrity formation. Presenting academic integrity as a topic to be taught early and explicitly — grounded, as this lecture has tried to model, in Ephesians 4:25's "members of one another" and Matthew 6:4's account of secret action, rather than only announced as a policy with penalties attached — treats the underlying issue as formational rather than merely regulatory, consistent with this chapter's overall argument that formation, not information, is theological education's actual center. Its limitation is the plainest one: teaching does not guarantee compliance, and a curriculum still needs an actual, fairly enforced policy alongside the teaching, not instead of it.

No single response here closes every gap, and this lecture does not claim otherwise; a seminary implementing several of these together, aware that none is sufficient alone, is closer to a realistic response than a seminary hoping a single detection tool or a single policy sentence will settle a problem this structural.