For Arts and Culture
Cognitive Sovereignty Checklist for Arts and Culture
When you use Midjourney, DALL-E, or ChatGPT in your creative work, the boundary between your artistic voice and the algorithm's suggestions blurs. When you curate based on AI audience data or write grant applications with AI assistance, you risk letting efficiency replace the genuine judgement that distinguishes meaningful cultural work. These tools can speed up your process, but they can also hollow out the very thing that makes your work worth funding or showing.
These are suggestions. Take what fits, leave the rest.
0 / 20 applicable
Tap once to check, again to mark N/A, again to reset.
Protecting your artistic practice and authorship
Document which parts of your work came from your own thinking and which came from AI generationbeginner
Keep a record of your initial concept, your sketches or notes, and exactly where you used Midjourney, Runway ML, or other tools. This helps you stay aware of what you actually made versus what the tool made. It also protects you if questions about authorship come up later.
Set a specific creative rule for when you will and will not use AI tools in a single piecebeginner
Decide in advance whether AI generates the base image and you paint over it, or whether you use it only for exploration before making final work by hand. Clear rules keep you from drifting into AI-assisted work without noticing.
Create one piece each month using zero AI assistance and reflect on what felt differentintermediate
Making work without algorithmic suggestions reminds you what your unassisted visual or conceptual choices sound like. This helps you notice when AI tools are pulling your work toward patterns rather than your own direction.
Ask a trusted peer to identify which elements in your work came from AI and which came from youintermediate
If another artist can spot the AI-generated parts, your audience likely can too. This reveals whether AI is enhancing your voice or replacing it.
Write a statement about your use of AI tools for each work and include it with your artist statementbeginner
Be explicit about whether you used AI to generate, edit, explore, or finish. Transparency protects your integrity and helps galleries, curators, and audiences understand what you actually did.
Experiment with AI tools on a separate practice project before using them in work you plan to showintermediate
Use ChatGPT or Midjourney as a sketchbook, not a shortcut. Understand how the tool works and what it tends to push toward before you let it influence work that carries your name.
Keeping curation and programming decisions genuinely yours
Before you look at AI audience data, choose three pieces or artists you believe deserve to be shownbeginner
Trust your curatorial eye first. Then check whether AI metrics support or contradict your judgement. If they strongly contradict it, dig into why before you change your mind.
Track which shows or acquisitions were shaped heavily by AI recommendations versus your own thinkingintermediate
At the end of each year, review your programming or collecting decisions. Did you choose work that performs well on metrics or work that challenges your audience? The pattern reveals how much the algorithm is steering your curatorial voice.
Visit three exhibitions or art spaces that explicitly reject AI audience analytics in their curationbeginner
See what curatorial choices look like when they are not optimised for engagement. Notice what gets shown, what tone the space has, and whether the work seems more experimental or narrower.
Write down your reasons for choosing each artist or piece in language a funding body will understand, before generating AI-assisted descriptionsintermediate
Use AI to refine your words, not to generate your reasoning. Your genuine curatorial argument is what distinguishes your taste and attracts the right audience.
Set a quota for showing work that performs poorly on audience metrics but meets your curatorial visionadvanced
Commit to including a percentage of artists or pieces chosen purely on merit, regardless of what analytics predict about attendance or engagement. This protects experimental work from being squeezed out.
Ask your board or stakeholders whether they want you to optimise for engagement or for cultural impact, and get them to chooseadvanced
AI metrics are built to optimise for one thing or the other. Make sure your institution has chosen what it actually values before you hand decisions to the algorithm.
Create a curatorial policy that names how you use audience data and where you override itintermediate
Write down the rules. You will use demographic data on who attends but not let it determine which artists you programme. You will track which shows sell tickets but reserve space for experimental work that may not. Published policy holds you accountable.
Keeping your voice in funding applications and institutional writing
Write the first draft of every grant application by hand or voice memo before you touch any AI toolbeginner
Your unfiltered ideas are more distinctive than what ChatGPT will generate. Capture your authentic argument first, then use AI to polish phrasing rather than to build the case.
Compare a grant application written with heavy AI assistance to one written mostly by you, side by sidebeginner
Read both to your colleagues and ask which one sounds like your organisation or artistic practice. The AI-heavy version will likely sound more polished but less like you.
Ban ChatGPT from writing sentences that explain why your work or organisation existsbeginner
Use AI to check grammar and refine wording. Use yourself to explain your mission, your artistic vision, and why you matter. That is the part of an application that moves a funder.
Set a rule that you will rewrite any AI-generated sentence in your application if you cannot explain it without looking at the screenintermediate
If you cannot say your own words out loud, you do not own them. This forces you to keep your voice even when you use AI as a writing tool.
Ask three people from your field to read your application and flag any sentences that sound generic or could apply to any organisationintermediate
AI-assisted writing tends toward the safe and general. Peers can spot where your application lacks personality before you submit it.
Build a sentence bank of phrases that describe your specific practice, and refuse to let ChatGPT rewrite themadvanced
Collect language that only you use to describe your work. Keep it. When AI tries to make it sound more professional or grant-friendly, push back.
Have your organisation adopt a policy on when AI can and cannot be used in funding applicationsadvanced
Some grant bodies will require you to declare AI use. Get ahead of this by deciding whether your organisation uses ChatGPT for checking tone and structure only, or whether it is off limits entirely for certain parts of applications.
Five things worth remembering
- If you use Midjourney or DALL-E, keep the original prompt and the unedited output alongside your final work. This trail helps you see how much you actually transformed what the algorithm made.
- Before you adopt any AI metric as a curatorial guide, ask yourself whether it measures what you value or what is easiest to measure. Engagement numbers are easy to measure. Cultural risk-taking is not.
- When a grant application using ChatGPT passes muster but yours written by hand gets rejected, do not assume you need AI. Ask the funder for specific feedback instead. They may want exactly what your unassisted voice offers.
- Curators and arts administrators: your judgment is your most valuable professional asset. AI tools get cheaper and faster every month. Your taste does not. Protect it.
- Cultural homogenisation happens quietly. If every gallery uses the same audience analytics software and every artist uses the same AI tools with the same settings, the work they produce will drift toward similarity without anyone deciding to make that happen.
Related reads
Common questions
Should arts and cultures document which parts of your work came from your own thinking and which came from ai generation?
Keep a record of your initial concept, your sketches or notes, and exactly where you used Midjourney, Runway ML, or other tools. This helps you stay aware of what you actually made versus what the tool made. It also protects you if questions about authorship come up later.
Should arts and cultures set a specific creative rule for when you will and will not use ai tools in a single piece?
Decide in advance whether AI generates the base image and you paint over it, or whether you use it only for exploration before making final work by hand. Clear rules keep you from drifting into AI-assisted work without noticing.
Should arts and cultures create one piece each month using zero ai assistance and reflect on what felt different?
Making work without algorithmic suggestions reminds you what your unassisted visual or conceptual choices sound like. This helps you notice when AI tools are pulling your work toward patterns rather than your own direction.