Public Speaking in the Age of AI (Part 2)

Preaching, formation and responsible practice

For Christian preachers as well as others whose work frequently involves public communication, AI offers considerable possibilities. A sermon is not simply religious information arranged persuasively, but preaching has always drawn upon tools of interpretation, theology, rhetoric and communication. Generative AI can become another such tool, assisting with biblical background, interpretive possibilities, theological comparison, sermon structures, illustrations and critical testing.

Part One of this two-part article argued that AI does not make human authorship obsolete. Instead, it relocates much of its distinctive value in judgment, interpretation, voice, presence and responsibility. The practical consequence is significant. Speechwriters and preachers should not stand at a distance from AI technologies, but engage them deliberately and intelligently as part of their ordinary professional practice.

The important question is therefore no longer whether AI should be used, but how it can be used well. At its best, AI can widen intellectual horizons, strengthen research, improve structure, sharpen prose, expose weak arguments and encourage experimentation. Used actively rather than passively, it can become a powerful research assistant, critic, tutor and editorial collaborator.

Preaching emerges from an interaction among Scripture, theological tradition, prayer, pastoral knowledge, contemporary experience and the particular congregation. AI can enrich several dimensions of this process. It can identify neglected themes, compare interpretations, raise exegetical questions and provide historical or cultural background. It may also help preachers move beyond habitual readings by exposing them to perspectives they would otherwise overlook.

Personal engagement with Scripture nevertheless remains fundamental. AI is most fruitful when it joins, rather than replaces, attentive reading. A preacher might first encounter the text directly and then ask AI what questions have been missed, which interpretations deserve consideration, what historical issues matter and where a proposed reading may be weak.

Used in this way, AI becomes a genuine interlocutor in interpretation. It can broaden the preacher’s field of vision while leaving theological discernment and pastoral application in human hands. Rather than threatening the craft of preaching, it may help preachers become more widely informed, critically self-aware and deliberate in their interpretive choices.

The most consequential question raised by AI concerns not only the texts produced but the formation of those who use it. Writing is one of the ways people learn to think. Searching for the right word, discovering that an argument does not cohere, reorganising paragraphs and rewriting inadequate conclusions can all deepen understanding.

This does not mean writers should avoid AI in order to preserve difficulty for its own sake. Rather, they should use it in ways that enhance learning. AI can explain why an argument is weak, compare alternative formulations, demonstrate possible structures and provide immediate feedback that would otherwise be difficult to obtain. Properly used, it can function as tutor, critic and editorial collaborator.

The same applies to preaching. Wrestling with a difficult text remains formative, but AI can enrich that process by introducing interpretive alternatives, challenging assumptions and identifying theological tensions. The danger lies not in extensive use but in passive use. Active collaboration, in which the preacher questions, tests, revises and sometimes rejects what the system proposes, can strengthen rather than weaken intellectual formation.

The deepest question is therefore not whether preachers and speechwriters should engage fully with AI, but what kind of practitioners they will become through that engagement. The aim should be thoughtful mastery rather than either avoidance or dependence.

One of the most valuable uses of AI is as a critical interlocutor. Writers can ask it to identify weaknesses in an argument, formulate strong counterarguments, expose unsupported assumptions, identify missing perspectives or suggest evidence that might challenge a conclusion.

Questions such as “What am I not seeing?”, “Whose perspective is absent?”, “Which assumptions remain unexamined?” and “What evidence would weaken this conclusion?” can materially improve a speech or sermon. AI is particularly valuable because such questioning can be repeated rapidly and from multiple perspectives.

It can also facilitate intellectual experimentation. Writers can compare structures, test alternative openings and conclusions, explore possible objections or generate several versions of a difficult paragraph. This capacity can enlarge intellectual agency by enabling writers to consider possibilities that limitations of time, experience or expertise might otherwise exclude.

AI systems do not approach questions from a neutral standpoint. Their outputs are shaped by training data, model design, system instructions and the linguistic and cultural patterns embedded within available sources. Even accurate answers may disproportionately reflect perspectives that are widely represented in digital material.

For preachers, this can mean familiar interpretations, dominant theological traditions or English-language scholarship receiving greater prominence than less digitised perspectives. Public speechwriters face similar risks when prevailing cultural or political assumptions are reproduced without sufficient scrutiny.

AI can also help address this limitation. Writers can deliberately ask for neglected perspectives, minority interpretations, non-Western scholarship or strong critiques of dominant positions. The responsibility for intellectual breadth remains human, but AI can make that breadth easier to pursue. Critical engagement is therefore preferable to avoidance.

Generative AI makes novel combinations of words easy to produce, but novelty is not identical with originality. Genuine originality often consists in fresh perception joined to responsible judgment. A good sermon need not say something no Christian has ever said before; it may enable familiar Scripture to be heard afresh in a particular setting. A good speech may reveal the significance of an established principle for a new moment.

AI can contribute to this process by helping writers test metaphors, compare formulations and explore unexpected connections. Yet originality ultimately depends upon attention to the actual situation, audience and purpose.

Specificity remains crucial. Good speeches and sermons require particulars: a person, place, event, object, biblical phrase, verified statistic, historical detail or observed gesture. AI can help locate or shape such material, but writers should prefer concrete reality to generic abstraction wherever possible.

The authority conveyed by fluent prose makes factual verification indispensable. Names, dates, statistics, quotations, historical claims and anecdotes should be checked against reliable sources. Biblical quotations should be verified against the nominated translation, while theological quotations should be traced to original sources wherever possible.

AI can make verification easier by identifying claims requiring checking, locating likely sources and exposing inconsistencies. It should nevertheless not be treated as the sole authority verifying its own output. The speaker remains responsible for accuracy.

Confidentiality also requires care. Pastoral conversations, counselling material, personal correspondence, personnel matters and unpublished organisational information should not be entered casually into AI systems. This is not an argument against extensive AI use, but for understanding the privacy settings, institutional policies and appropriate boundaries of the tools being used.

Transparency likewise calls for judgment. AI use ranges from grammar correction and research assistance to structural revision and substantial drafting. A reasonable principle is that the greater AI’s substantive contribution, the stronger the case for appropriate disclosure where professional or institutional expectations make that relevant. Thoughtful transparency can strengthen rather than weaken trust.

AI can suggest what might be said, but the speaker remains responsible for saying it. This distinction matters especially when addressing grief, injustice, political controversy, conflict or trauma. A sentence may be accurate and rhetorically effective while nevertheless being pastorally unwise.

Writers should therefore ask what their words are likely to do to those who hear them, who may experience them differently, and whether the occasion justifies the claim being made. AI can help anticipate such reactions by simulating alternative audience perspectives and identifying potentially insensitive language. Used in this way, it can strengthen ethical reflection.

Responsible communication seeks understanding without manipulation and persuasion without deception. AI should therefore be judged not merely by whether it makes communication more efficient, but by whether it helps produce speech that is truthful, humane and worthy of trust.

A mature workflow should integrate AI throughout the writing process while retaining human agency. The process begins with audience, occasion, central proposition and desired response. For preaching, it includes attentive engagement with Scripture; for public speechwriting, careful engagement with relevant evidence and context.

AI can then assist with research, background, structure, alternative interpretations, drafting, compression, counterarguments and stylistic refinement. It is especially valuable for identifying omissions, testing transitions, challenging assumptions and locating points requiring verification. Human judgment then determines what should remain, what should be revised and what should be rejected.

The final stage should be oral. The complete speech or sermon should be read aloud with attention to rhythm, clarity, length, repetition and naturalness. The speaker must finally ask: “Are these words I understand, believe and am prepared personally to own?” That question establishes the proper relationship between technological assistance and human responsibility.

The argument of both articles can now be stated positively. The goal should be neither AI-free writing nor AI-dependent writing. It should be AI-assisted human authorship in which technology enlarges attention, wisdom, craft and responsibility. AI can expose writers to objections they had not considered, compare interpretations, improve structure, sharpen prose, stimulate creativity and relieve routine labour. Used confidently and critically, it can free time for deeper research, reflection, prayer, conversation and engagement with people.

As machines become better at generating language, humans have an opportunity to become better at questioning, interpreting, discerning and editing. The distinctive contribution of the speechwriter or preacher lies not in competing with AI at sentence production, but in determining what is worth saying, what is true, what the occasion requires, how particular people should be addressed and which words the speaker is willing to stand behind.

The deepest question is not whether AI can produce an impressive speech or sermon but how speechwriters and preachers can use increasingly powerful tools to become more thoughtful, informed, creative and responsible communicators. Good communication has never consisted merely in producing fluent sentences. At its best, speechwriting and preaching involve disciplined attention to truth, people, context and purpose. AI can become an increasingly valuable partner in that work, provided it is engaged confidently, critically and with a clear sense of human agency and responsibility.


Rev Dr Rod Benson is General Secretary of the NSW Ecumenical Council and a minister of the Uniting Church in Australia serving at North Rocks Community Church in Sydney.

To read Part 1 of this article, click here.

Image source: author

Leave a comment