When AI becomes a crutch
Generative artificial intelligence offers remarkable possibilities for preaching, speechwriting and other forms of public communication. It can assist with research, summarise complex material, propose structures, identify weaknesses in arguments, generate counterarguments, suggest illustrations and improve prose. Used judiciously, it can function as research assistant, editor, interlocutor and critic.
Risks for novices
Yet the capacities that make generative AI useful also create significant risks for novices and for those who see the technology as a way of cutting corners in their most important work. AI can perform many of the intellectual and creative tasks that have traditionally contributed to the formation of preachers, writers and speakers. The central question is what kinds of use enhance human competence and what kinds circumvent the processes through which competence is acquired.
This distinction is important when comparing experienced practitioners with those still developing their craft. A preacher or speechwriter with substantial knowledge and experience may possess the critical resources necessary to assess, correct and improve AI-generated material. They recognise an implausible interpretation, unsupported factual assertion, rhetorical cliché or superficially persuasive argument. A novice will likely lack those resources. For an experienced practitioner, AI may extend capacities developed through education and practice; for an inexperienced practitioner, it may substitute for capacities not yet acquired.
The less secure the underlying competence, the more cautiously the relevant intellectual task should be delegated. A theological student learning the skills of biblical exegesis may need to independently perform tasks that an experienced biblical scholar can reasonably automate. Similarly, a novice speechwriter needs to learn how evidence becomes argument, and how rhetoric works in practice, before routinely asking AI to construct arguments and write persuasive professional prose.
I see three key risks associated with the use of generative AI for sermon preparation.
1. Absence of foundational skills
It’s the familiar chicken-and-egg dilemma. Before a communicator can critically evaluate what AI produces, they need sufficient knowledge to recognise whether its output is accurate, appropriate and persuasive. In preaching, this includes competence in biblical interpretation, hermeneutics, theology, Christian tradition, ethics, pastoral practice and homiletics.
These abilities develop through sustained practice. Students learn exegesis partly by doing exegesis; preachers learn to construct sermons by constructing sermons; writers learn argument by wrestling with evidence, ambiguity and counterargument. Failure, correction and revision are not merely interruptions to this process. They are among the means by which judgment develops.
Generative AI greatly expands the possibility of cognitive offloading: transferring part of an intellectual task to an external system. Such offloading is not inherently undesirable. Human beings have always extended their cognitive capacities through writing, books, libraries, calculators and computers. The significant question is which tasks should be delegated, at what stage of formation, and for what purpose. There is a difference between using a calculator after understanding arithmetic and using one instead of learning arithmetic. Something analogous applies to exegesis, theological reasoning and rhetorical composition.
A preacher who asks AI to interpret a biblical passage before carefully reading it may miss the textual difficulty that would otherwise provoke deeper study. A theological student who immediately requests several interpretations may receive plausible alternatives without learning why interpreters disagree or how competing interpretations should be evaluated. A speechwriter who asks AI to formulate an argument before examining the evidence may acquire a coherent structure without developing the analytical ability required to construct one. The problem is not simply that AI may be wrong. Fluent output can conceal the extent to which judgment has been outsourced.
One practical response is therefore to undertake the first stage of significant intellectual work independently: read before prompting, investigate before summarising, formulate the problem before requesting solutions, and develop a provisional argument before asking AI for alternatives. AI can then become a critical interlocutor rather than a substitute thinker.
2. Truth, verification and responsibility
Generative AI can produce inaccurate factual claims, misleading summaries, erroneous quotations and fabricated or incorrect references. Tools combining language models with search and retrieval can reduce some forms of error, but linguistic fluency should never be confused with reliability. A proposition does not become true because it has been expressed confidently and elegantly.
The deeper ethical issue here is epistemic responsibility. A preacher, lecturer, advocate or political speaker cannot reasonably defend an inaccurate statement by blaming AI. Public communicators remain responsible for claims made under their names. Significant factual assertions should therefore be rigorously verified, quotations checked against original sources where practicable, and important references independently confirmed. Where matters are disputed, uncertainty should be acknowledged.
This responsibility is especially important in preaching. Biblical interpretation inevitably involves advanced skills of judgment, and theological traditions frequently disagree about the meaning and application of texts. AI can map interpretive possibilities and identify relevant scholarship, but it cannot relieve the preacher of responsibility for determining what should be proclaimed. AI may assist the investigation, but responsibility for truth claim remains human.
3. Weakening of agency, creativity and voice
A further risk concerns authorship and creative agency. Good public communication does more than transmit information. It communicates the intellectual, moral and imaginative agency of a particular person. Experienced communicators develop recognisable habits of thought and expression: characteristic rhythms, vocabulary, metaphors, humour, theological emphases and methods of argument. These qualities develop gradually through reading, listening, writing, speaking and revision.
Writing itself is a mode of thinking. Writing with a pen or pencil on actual paper enhances mental processes and sharpens thinking. In an earlier career, I taught my theological students to avoid screen-based technologies where possible when conducting basic academic research. We frequently discover what we believe, recognise weaknesses in our reasoning, and find connections between ideas through the struggle to articulate them. This is more natural when limited to pen and paper. The difficulty of composition is often not an inefficiency to be eliminated but the process through which clearer thought emerges.
Generative AI changes this experience because it can supply polished language almost instantaneously. Faced with a blank page, the writer can obtain an introduction, outline or complete draft within seconds. This may improve efficiency, but efficiency is not the only good involved in intellectual work. The stubborn paragraph, unresolved argument, ambiguous biblical text or elusive metaphor may compel the communicator to think more carefully. Some apparently inefficient processes are formative.
There is also a plausible risk of homogenisation. Generative AI is highly effective at producing orderly, conventional and stylistically competent prose. When users accept such outputs with limited revision, communication may converge toward predictable patterns. The danger is not that AI inevitably destroys originality, but that its convenience may reduce the incentive to pursue the particular, unexpected or idiosyncratic insight that gives communication its distinctive character.
The value of human initiative
The appropriate discipline is therefore to preserve human initiative: develop the central claim before asking AI to test it, construct an outline before requesting alternatives, draft important passages independently, and use AI to expose weaknesses rather than merely supply finished prose. Ask yourself: Is AI enlarging my agency, or replacing an activity that I ought to perform? What am I losing by resorting to the efficiency of AI in my sermon preparation?
At its best, AI can enlarge human capability by widening inquiry, testing assumptions, sharpening arguments and freeing time for reflection, creativity and human encounter. The goal is neither technological enthusiasm nor suspicion, but mature judgment about what to delegate and what to retain. Preachers and other public communicators need not fear these tools. Used critically and creatively, generative AI can become not a crutch that weakens formation but a valuable conversation partner that helps us become more discerning, imaginative and responsible communicators.
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.
Image source: author
All Told 