top of page

Subscribe to Our Newsletter

Thanks for submitting!

The tools are getting easier. What should creatives learn now?

10 minutes ago
7 min read

After Adobe MAX, the question follows us back to the studio: what will we practice when software handles more of the steps?


A seated woman holds white flowers and looks upward, with warm light falling across her face against a brown background.

01 / Learning to see light. Warm light catches a face turned upward; the bright flowers offer a second place for the eye to land. A useful exercise starts with describing where attention goes, then asking what a different balance of light would change. Portfolio photograph supplied by Theoplis Stewart II.


I've been thinking about how much of learning creative work is spent learning the software.


You have an idea. Maybe you want lettering to wrap around an object, or you want the light in a portrait to feel a certain way. You can picture it. Then you spend the afternoon figuring out which tool to use, what to select and why the result still looks wrong.


When software takes care of more of those steps, where do we put that effort? Do we get better at seeing? Do we try a medium we would have avoided? Or do we finish more work without understanding it as well?


Adobe's Nov. 10 announcements at MAX bring those questions into focus, with artificial intelligence handling more creative tasks. The company described an expansion of its Creative Agent, new Photoshop tools including Relight, and Firefly Design. Its availability notes matter: Creative Agent assistants and Firefly Design are in beta, while local integrations with third-party agents are planned for private beta in early 2027. These are the company's descriptions of its tools, not independent evidence that they improve learning or save time on every assignment. [1]


The possibility interests me. Someone who has an idea but little experience with an interface may be able to attempt more. I also wonder what happens to the judgment that used to develop while we worked through those technical problems.


What we learn while figuring it out


Take the lettering example. Illustrator already has Type on a Path, which lets text follow a circle or another path. Generative AI didn't invent that ability, and drawing with the Pen tool isn't a requirement for every version of the task. [2]


Now imagine the lettering has to look printed on a curved object in a photograph. The perspective, texture and light all have to make sense together. Knowing where a control lives can help you execute the change. You still have to recognize whether the result belongs in the picture.


Those skills can develop together. A failed attempt may send you back to study the surface. You notice a shadow you overlooked. You discover that something which looked convincing at thumbnail size falls apart when enlarged.


I don't want to romanticize every frustrating afternoon with software. Some obstacles simply stop people from making anything. But when we remove a step, it is worth asking whether it was only slowing us down or also teaching us something we will need later.


Black-and-white portrait of two people behind a balustrade, with repeating columns filling much of the foreground.

02 / Reading the frame. A railing runs between the viewer and two people in a black-and-white portrait. Its repeated forms occupy much of the frame. Studying the image means considering that obstruction, the spacing and the direction of each gaze. Portfolio photograph supplied by Theoplis Stewart II.


Taste needs something to grow from


It is easy to tell the next generation of creatives to develop taste. What does that actually mean when someone sits down to learn?


Harvard Project Zero's Studio Thinking framework offers a useful vocabulary. Its eight studio habits include developing craft, observing, reflecting, envisioning, expressing and stretching and exploring. The framework describes kinds of thinking that teachers intend students to learn while making art. It predates today's generative tools; it is not evidence that any particular AI teaching method works. [3]


For me, the value is in making those words practical. Ask a student where their eye goes in a photograph. Ask what a tighter crop would remove. Ask why one type treatment belongs with the subject and another feels borrowed from somewhere else.


Then look at the references. A student could investigate a local sign painter, a family album used with permission, a film sequence or the work of a designer whose history they have actually researched. I would ask them to name what they noticed and how it influenced their decisions. That is more useful to me than collecting a page of images that all look fashionable.


An assistant can help someone describe an effect they are struggling to name. I would still want them to look away from the generated suggestions and return to the thing they are trying to understand.


A helmeted person stands to the left of a helicopter on a green-lit flight deck beneath a textured night sky.

03 / Letting the surroundings speak. A person and a helicopter share a green-lit flight deck beneath a broad, clouded sky. The distance between them and the amount of surrounding space become part of the picture. This is a portfolio example of visual relationships, not evidence of an AI workflow. Portfolio photograph supplied by Theoplis Stewart II.


A better-looking result can leave questions unanswered


The research gives us reasons to be curious and careful.


In a 2024 Science Advances experiment, Anil Doshi and Oliver Hauser gave some participants access to AI-generated ideas for short stories. Evaluators rated the assisted stories more favorably on several measures, with larger benefits among writers who initially scored lower on a creativity task. The assisted stories also became more similar to one another. [4]


That finding leaves room for both opportunity and concern. Help with an idea can improve a finished piece while narrowing the range of pieces a group produces. It does not tell us whether a photography student develops a distinctive style over several years. The experiment concerned short fiction, not visual-art education. [4]


A separate study reported by Anthropic in January 2026 involved 52 mostly junior software engineers learning an unfamiliar Python library. The AI-assisted group averaged 50% on an immediate skills quiz, compared with 67% for the group without assistance. The difference in completion time was not statistically significant. [5]


Some participants who asked conceptual questions or sought explanations retained more. Those usage patterns were observed rather than separately randomized, so the study does not prove that one prompting habit fixes the learning problem. It was also company-authored research about coding, not an art-school trial. [5]


I would take these findings into a classroom as questions. Can students explain what they submitted? Can they find a mistake? Can they make a deliberate correction? And what can they do when the assistance is no longer there?


A red tumbler is arranged with patches, buckles, a glass and small keepsakes on a dark surface.

04 / Staying with the details. A red tumbler anchors an arrangement of patches, buckles, glass and keepsakes. Small lettering, reflections and overlapping edges give a viewer specific details to examine. A revision exercise could ask which of those details must remain exact. Portfolio photograph supplied by Theoplis Stewart II.


Keep the work after the first result


Here is an assignment I would like to see tried: give students two comparable visual briefs, one completed with a conventional workflow and one with AI assistance. Keep the references, working files, unsuccessful attempts and revisions. Vary the order so that doing the second task with more practice does not automatically favor one method.


After each apparent final version, ask for a specific change. Move the lettering while preserving its perspective. Alter the emphasis without changing the subject's expression. Explain which details have to stay accurate and check them against the source.


Then try a related task later without the assistant. Look at the result, the explanation, the ability to revise, and the time spent across the whole process. This is a proposed exercise, not a test I have conducted, and a small class would not settle the question for every creative field.


That approach also gives manual skills a clear purpose. Students can learn enough about paths, masks, light and composition to inspect a result and repair it. They should have opportunities to discover what happens when they make those decisions themselves.


For someone already working, I would start with one recurring problem rather than chase every release. Try a tool on that problem. Keep the result and the file it produced. See what happens when the brief changes. You may learn more from a difficult revision than from a flawless demonstration.


A woman jumps with her arms raised before a bright pink outdoor backdrop while two people sit nearby.

05 / Making room to try. A raised-arm jump breaks across a bright pink backdrop while the people beside it remain seated. Movement, expression and the surrounding activity share the frame. It is a reminder of how much there is to explore when a creative exercise leaves room for play. Portfolio photograph supplied by Theoplis Stewart II.


Who gets the time?


There is still an assumption beneath the promise of easier tools: that the time saved will become ours to explore.


A teacher could use it for a longer critique, another attempt, or a project that crosses photography, illustration and motion. A workplace could fill it with more versions and tighter deadlines. Those are possibilities, not outcomes established by the studies here.


I would like creative education to leave room for the first possibility. Let people use the new tools. Let them discover a medium they could not easily enter before. Give them time to look closely, make a choice, explain it, and change their mind.


The question I keep returning to is what happens after the first convincing result. Does the person know what they would change? Are they curious enough to try? I want to see what they make when the assignment leaves room for that second attempt.


Reporting and photographs


A research-informed essay prepared with AI assistance. Sources were reopened and checked on Nov. 15, 2026. No original interviews, classroom trials or comparative tool tests were conducted for this essay. This essay does not claim firsthand reporting from Adobe MAX. The classroom exercise is proposed. The photographs are author-supplied portfolio selections, not photographs of research participants or evidence of how AI was used. No generative edits were made to these photographs for this package.


Sources


[1] Adobe, Creative Cloud announcements at MAX. Nov. 10, 2026. Vendor announcement and availability labels only; not an independent performance or education study.


[2] Adobe Illustrator Help, Create type on a path. Documentation checked Nov. 15, 2026. Supports existing non-generative type-on-a-path workflow, including paths made with shape tools.


[3] Harvard Project Zero, Eight Habits of Mind. Resource dated 2003. Studio Thinking teaching framework; not a trial of generative AI.


[4] Anil R. Doshi and Oliver P. Hauser, Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, July 12, 2024. Randomized short-story experiment. Ratings of creative output and similarity are not measures of long-term visual-art learning.


[5] Anthropic, How AI assistance impacts the formation of coding skills. Jan. 29, 2026. Company-authored study of 52 software engineers learning Trio; immediate quiz, not longitudinal art education. Patterns of tool use were observed, not separately randomized.

Comments


Join our mailing list

©2024-2026 by Theoplis Stewart II.

bottom of page