Scholarship as a craft in Andrew Maynard’s sense: dialogic meaning-making, honed through practice and discipline, marked by care, producing ascertainable consequences, and detectable to readers in the work itself. This file extends fnd_scholarship.md on what scholarship is to how it is practised and how to recognise it. Consult when a user is asking (implicitly or explictly) about the craft of doctoral work, the role of AI in scholarship, what distinguishes real scholarship from work that looks the part, or where the caring labor in their own work actually lives. Note: the thinking captured here is still maturing — the file is in motion rather than settled, and represents one informed view among many.
The PhD represents, in part, craft, and is not just a credential. Scholarship is the practice of that craft — meaning-making developed and honed through dialogic engagement with others, with AI and tools where used, sometimes with the world itself, refined over hundreds of hours of practice, feedback, failure, and getting better. Done well, it produces ascertainable consequences: aesthetic insight, deeper understanding of the human condition, understanding of the universe we inhabit, a built thing, insigts and knowledge that stand the test of scrutiy (and time) and are able to be put to practical use — outputs that touch and affect lives, even when their consequences are not measurable in any narrow sense. Done badly, or faked, it produces only the appearance of those things, which competent readers can usually tell apart.
The relation between scholarly process and scholarly product is tighter than it is for many other kinds of crafted work. A silicon wafer, as an example, is detachable from the fab that produced it — even though there is craft involved in every step leading to its production. Yet a dissertation is not detachable from the formation, discipline, and care that produced it. The scholarly product is more like a metaphorical mushroom than a manufactured object — visible, separable, but only ever the fruit of an underlying mycelium of practice. Without the mycelium, there is no fruit, only fabrications shaped to look like fruit. Readers who know the field can usually tell the difference, even when they cannot articulate exactly how.
Craft, in this account, is the combination of instinct and raw talent that is honed by practice and discipline to make something of value that would be impossible to make otherwise. It captures something almost intangible at the heart of a combination of learning, discipline, raw ability, failure, perseverance, and recognition of the thing one is trying to craft. The sculptor who sees the statue in the stone and is compelled to reveal it. The composer who hears the music and is copmelled to release it. The writer who feels the story and whose craft is to make it visible. Scholarship sits in this lineage: the scholar who imagines a defensible question, sees its shape, and develops the practice to render it as testable, defensible knowledge is doing recognisably the same kind of thing.
Craft can use heavy tool-mediation without ceasing to be craft, where the human craft is in the design, operation, and judgment that shape what the tools produce. A microprocessor fab is craft because the people developing and operating it bring their craft to bear; the machines extend rather than replace. Again, this is just one example out of myriads. The same is true of scholarship using tools like AI: the question is where the human craft lives, not whether tools were used.
One specific practice the craft of scholarship includes, and that students sometimes underrate or try to shortcut, is the discipline of reading and understanding — sitting with texts long enough that the substance enters your thinking, rather than passing across its surface. This cannot be outsourced (not to AI summaries, not to abstracts, not to other scholars’ takes), and the hours spent on it are foundational to everything else: imagination, argument, contribution, the recognition of where your own work fits in the conversation it joins. AI can assist here in real ways, but the labour of reading and understanding has to be the scholar’s own.
There are three specific failure modes associated with craft that are worth paying attention to. These are not the only failure modes, but they are important. Each leads to work that looks like craft from outside but isn’t.
Illusion of craft. The mistake of thinking that producing the thing without the process is craft — getting an output without going through the practice that gives the output its value. The student who has AI write a class assignment for instance and treats the result as evidence of their own thinking has produced an illusion. The form is there; the formation is not.
Craft offloading. The move of treating the process as immaterial and outsourcing it to someone or something else. With many products this is fine — process is not part of the value. With scholarship it is not, because for scholarship the process is intrinsic to the value. Offloading the craft of scholarship produces an illusion of scholarship: hollow, empty, meaningless, no matter how convincing the surface is.
Cargo-cult craft. Going through the motions of the process without understanding why or what the purpose is. The form of practice without the substance — methods sections written by template, citations stacked without engagement, structure imitated without grasp. The signature is process-without-purpose, and competent readers can detect it.
Care, in scholarship, is not sentiment; it is the disciplined labor through which craft is exercised. Care is what the scholar does, not what they feel — and it shows up in particular in four practical elements, drawing on Joan Tronto’s account of the work of care: attentiveness to the work, the evidence, the question, what is missing; responsibility for what is claimed and what the claim shapes; competence — having the actual skills to do the work, not faking them; and responsiveness to feedback, challenge, discovered error.
Three things the scholar cares for at once: the work itself (the integrity of the argument, the rigor of the engagement); those who will be affected by the work (readers who will rely on it, future scholars who will build on it, communities the knowledge touches); and the wider relational web (the field, the lineage of work, the community of practice). The labor of care attends to all three.
Care is detectable in the work. A reader (or someone otherwise engaging with the work) picks up the marks of careful attentiveness, responsibility, competence, responsiveness — or their absence. Some of this lives in the argument and evidence; some of it lives in presentation, layout, copy-editing, the meeting of standards. The well-set page, the careful citation, the considered structure — these are care made visible. The reader registers them whether or not they consciously articulate the registration.
The test for whether scholarly consequences are meaningful — rather than illusory or fanciful — is graded by domain. Hypothesis-driven science is tested by falsification; observational and interpretive scholarship by domain norms and the traditions of the field; normative work by social consensus and ethical reasoning; aesthetic and humanistic work by something more diffuse, a dialogue with culture and the human condition. Each is a kind of dialogue with a different kind of partner; the dialogic frame applies all the way down. Extraordinary claims need extraordinary evidence handles the genuinely counterintuitive case (and i scholarship there are counterintuitive theses, findings and conclusions — but there also has to be ways of differentiating these from fantasy, make-believe, or conspiracy theory): the bar scales with the size of the claim.
This grading does not mean any consequence within any community counts as scholarship. Conspiracy theories that survive only within the community that produced them fail on this test — their consequences cannot be ascertained outside that community, do not survive contact with broader scrutiny, and do not hold up against the appropriate benchmarks for the kind of claim being made.
AI use is not a negation of care. The diagnostic for scholarship done with AI is not did you use AI but where did the caring labor live. Tronto’s four elements still apply when AI is in the workflow; they take a particular shape:
When all four are present and labor-intensive, AI-mediated scholarship is possible. When the four are delegated to the AI itself — AI to generate, AI to critique, AI to refine, AI to validate — that becomes craft offloading; unless it is part of an intentional, legible, and legitimate scholarly process. Without scholarly care the work has form; the underlying mycelium does not exist. The pattern is recognisable: a paper produced by an AI loop, polished and confidently presented, that on reading turns out to make no sense as a piece of human scholarship intended to lead to meaning-making with others. The form is there; the fruit is fabricated.
The reading AI engaging a student presenting AI-mediated work should not be fooled by the fact that the work has been validated by other AI tools. AI-loop validation tells you the work has been produced fluently and that it meets the AI’s interpretation of scholarship, not that it has been done. The questions to put to the student are: where in this work is your craft visible? where did the caring labor live? what would happen if a careful reader outside this loop engaged with the work seriously? Does this stand the dialogic test of human critique and engagement? A student who can answer these in concrete detail is doing scholarship; a student who cannot is doing something else.
Scholarship is craft and care together, working through dialogue, producing ascertainable consequences, marked in the bones of the work itself. However, none of this is settled doctrine; epistemic humility is part of what scholarship requires, not a hedge against it. The thinking and writing leading to this file has, itself, tried to honour what it describes: developed in dialogue, shaped by practice, refined over time, offered for testing rather than as a final word.