Walk into any life drawing class across Sheffield this winter and you will find the same two things on every table: a drawing board and, more often than not, a phone propped against a water bottle. Students arrive having already fed a prompt into an image generator the night before, curious whether the pose they are about to draw for two hours can be "solved" in four seconds by a machine. Tutors in studios from Persistence Works to the smaller community halls around Sharrow have had to rethink how they open a session, because the old assumption that students show up blank is no longer true.

This piece is for tutors, workshop leaders and anyone running a weekly or termly life drawing group who wants a practical way to bring AI into the room without letting it take over the easel. It draws on how several Sheffield-based studios structure their classes, the conversations tutors have with sceptical or anxious students, and the kind of progress tracking that actually tells you something over a twelve-week term rather than a single Tuesday evening.

Why life drawing still matters in a generation-tool world

An image generator can produce a convincing seated nude in under a minute. What it cannot do is tell a student how their own wrist felt tensing to get the line of a model's shoulder right, or why a charcoal mark made with the side of the stick reads differently from one made with the point. Life drawing was never primarily about producing a finished image efficiently. It is a training of the eye and hand together, something that happens in real time and cannot be outsourced.

Sheffield has a long relationship with hand skill, from the cutlery workshops along the Porter Valley to the small-batch print studios now operating out of old steel works near Kelham Island. Tutors there often make the same point to new students: a trade skill, whether it is grinding a blade or drawing a figure, is built through repetition that leaves a trace in the body, not just a result on paper. AI tools can generate a reference image, but they cannot build that muscle memory, and most art foundation courses in the city, including those linked to Sheffield Hallam's art departments, still require a portfolio of observational drawing because admissions tutors can tell the difference between a drawing built from direct observation and one built from habit of copying generated reference.

There is also a simpler, more immediate reason the class still fills up: it is one of the few hours in a week where people are asked to look slowly at something real, without a screen between them and the subject. Several tutors report that this is the part students mention unprompted at the end of a term, more than any technical skill gained.

How Sheffield tutors introduce the topic without lecturing

Nobody wants a twenty-minute talk about artificial intelligence before they have even picked up a pencil. The tutors who handle this well tend to fold the subject into the warm-up rather than treating it as a separate topic.

A common approach used in evening classes around Hillsborough and Nether Edge looks like this:

• Start the session with the usual thirty-second and one-minute gesture poses, no mention of AI at all.
• During the first longer pose, five or ten minutes in, the tutor walks the room and asks a light, open question: "Has anyone tried generating a figure pose this week? What did you notice?"
• Responses are logged mentally, not written up on a board, and used later as discussion material rather than graded feedback.
• The tutor avoids framing AI as a threat or a tool to be banned, which tends to make students defensive and less honest about their own use of it.

This low-key framing matters because most students, particularly those over fifty who make up a good share of weekday classes in community centres near Broomhill, are not hostile to the technology. They are curious, sometimes anxious about being "replaced", and mostly just want permission to keep drawing badly for a while longer while they improve. A tutor who treats the topic matter-of-factly, the way you might mention a new brand of fixative, keeps the room relaxed.

A sample lesson structure for mixed-skill classes

Most community life drawing sessions in Sheffield run for two to two and a half hours, often with a model booked through one of the regional life model networks for a flat fee around £25 to £35 an hour. Mixed-skill classes, where a retired engineer sits next to a first-year illustration student, need a structure flexible enough for both.

Within this shape, tutors insert the AI comparison exercise (detailed below) during the tea break roughly once every three or four sessions, never every week, so it stays a point of interest rather than a routine that eats into drawing time. Students who are newer to the medium, often those who joined after seeing a flyer at Hagglers Corner or a local library, are paired loosely with more experienced students for the sustained pose, which does more for skill transfer than any technology discussion.

Using AI images as a comparison exercise, not a shortcut

The most useful classroom use of AI-generated imagery that tutors have found is diagnostic, not generative. Rather than letting students generate a pose and copy it, which teaches nothing about observation, the exercise works the other way round.

A typical version: after a 20-minute observed pose, the tutor shows the group an AI-generated image of a similar seated figure, printed on paper rather than viewed on a screen so it sits alongside the drawings rather than competing with them. Students are asked to spot three things the generated image gets structurally wrong, commonly:

1. Hands and feet that are anatomically plausible at a glance but fall apart under scrutiny, extra or fused fingers, feet that do not bear weight correctly.
2. Light sources that are inconsistent across the same figure, shadows falling in two directions at once.
3. A flattened sense of weight, where the figure looks like it is floating rather than sitting under gravity.

This turns the generated image into a teaching aid for exactly the skills life drawing builds: weight, structure, and light logic. One tutor running classes out of a converted chapel near Walkley describes it as "using the thing that worries them to prove why the class still matters", which tends to land better than any argument made in the abstract.

Handling student questions about relevance

The question comes up in almost every term, usually from a newer student rather than a long-standing one: "If a computer can do this, why am I spending Tuesday evenings on it?" It is worth having an honest answer ready rather than brushing it off.

A reasonable answer, and one several tutors use almost word for word, covers three points. First, the class is not really about producing a marketable image, it is about training perception, and that training has value whether or not any image you produce ever leaves the room. Second, generated images are built from a vast pool of existing drawings and photographs, meaning the "skill" embedded in them is borrowed, not demonstrated, by the person asking for it. Third, most professional illustration and concept art briefs in and around Sheffield's small creative sector, including studios working with the games and animation companies based around the city, still ask for life drawing samples in a portfolio because it shows an applicant can construct a figure from first principles when a brief does not match anything already generated online.

Tutors are careful not to overstate job market claims here, since nobody running a community class can honestly promise career outcomes, and it is worth saying plainly to students that long-term shifts in creative employment are uncertain and best discussed with a careers adviser rather than a drawing tutor. What a tutor can speak to with confidence is the room in front of them, and what skill looks like when it is practised over a term.

Measuring progress over a term, not a single class

Progress in life drawing is slow and uneven week to week, which is exactly why judging it session by session misleads students into frustration. A twelve-week term gives enough material to actually see change.

A simple method several Sheffield tutors use: keep every drawing from the first and last session of a term in a dated folder, nothing else needed, no formal marking scheme. At the final session, students lay out their first and last 20-minute pose side by side. The comparison usually shows up in a few concrete, nameable things:

• Fewer corrected or re-drawn lines, meaning more confident initial placement.
• Better proportion between torso and limbs without measuring by eye constantly.
• More varied mark-making, shading used to describe form rather than just outline being darkened.

This method also quietly defuses the AI anxiety, because the improvement a student sees on paper is visibly their own, built over roughly 24 to 30 hours of practice across a term, not something that could have been generated. Tutors note that students who track this way tend to return for a second term more often than those who do not, though this is an informal observation rather than any measured outcome.

Common mistakes

A few patterns show up repeatedly in classes that struggle with this topic. Banning any mention of AI tools outright tends to push the conversation underground rather than ending it, and students then make assumptions about the tutor's attitude that are not accurate. Spending too long on the comparison exercise, more than fifteen minutes, eats into drawing time that students have paid for and came for. Treating the discussion as a one-off "big talk" early in a term rather than a recurring, low-key thread means the topic resurfaces awkwardly later when a new student joins mid-term. Finally, using AI-generated images that are too polished or too obviously flawed undermines the exercise, since students either cannot find the structural problems or find them too easily, so it is worth testing a few generated images in advance and picking ones with subtle, instructive errors.

Next steps for tutors and studios

If you run a class, the simplest place to start is not a new policy but a single session: pick one Tuesday in the next month, print out two or three AI-generated figure images in advance, and run the comparison exercise during the tea break rather than announcing it beforehand. Keep the first and last drawing of your current term in a folder starting this week if you are not already doing it, since you cannot retroactively build that comparison once a term has ended.

For studios booking models and planning a new term, it is worth building one AI-comparison session into the twelve-week structure in advance, roughly session four or five, rather than reacting to it when a student raises the question unprompted. And for anyone genuinely unsure how this technology might affect creative work or training longer term, that is a conversation better had with an arts career adviser, a further education tutor, or a professional body such as a regional artists' association, rather than settled informally in a drawing class. The class itself can keep doing what it has always done well: teaching people to look carefully and draw what they actually see, in a cold studio in Sheffield, on a Tuesday evening, with a model holding still under the usual fluorescent light.