40 Questions Arts and Culture Should Ask Before Trusting AI
When Midjourney generates an image, when ChatGPT drafts your funding application, when audience data shapes what you programme, you need to know what you are actually deciding about. These questions help you keep your judgement in charge.
These are suggestions. Use the ones that fit your situation.
1When you use Midjourney or DALL-E to generate visual work, can you articulate what human decisions shaped the output (your prompts, iterations, selections, combinations with other materials)?
2Does the work you have made with generative tools still reflect your specific artistic concerns, or has the tool's aesthetic preferences taken over your voice?
3If you are using Runway ML for video work, where does the machine learning's contribution end and your creative judgement begin in the final piece?
4When you credit your work, are you being honest about whether it was made with AI assistance, and does your audience have that information?
5Are you using AI tools to explore ideas you could not explore before, or are you using them because they are faster than your actual artistic process?
6If you removed the AI tool from your workflow, would the work still be recognisably yours?
7Does using Adobe Firefly or similar tools to accelerate production phases change what risks you are willing to take in your work?
8When you iterate with a generative tool, are you making intentional creative choices or accepting the first thing that looks acceptable?
9What artistic problems require human struggle and experimentation, and are you outsourcing those to AI when you should not be?
10If another artist used the exact same prompts in Midjourney or DALL-E, would they make the same work you did?
Curation and Programming Decisions
11When you review audience analytics that an AI system has flagged as high-engagement content, do you know what behaviour that metric actually measures?
12Has AI audience data ever recommended that you programme something you do not think is good, and did you do it anyway?
13Are you programming work based on what existing audiences already respond to, or are you risking work that introduces them to something new?
14If you ignored the AI recommendations about what to show, what would you programme instead, and why would that be the better choice?
15When AI systems analyse which artists or art forms drive engagement, are those systems missing whole categories of value your organisation actually cares about?
16Do the artists you think are most important get recommended by AI audience data, or does the algorithm favour something else?
17Has an AI system ever recommended against programming something because of predicted low engagement, and did you consider that recommendation seriously enough?
18What does your curation look like in five years if you keep following AI engagement recommendations?
19Are you curating for the audiences you have right now, or the audiences you want to develop?
20When you make a curation choice that contradicts AI recommendations, can you articulate why your human judgement was right?
Funding, Grants, and Applications
21If you used ChatGPT to draft or edit your funding application, does the final text still sound like your organisation, or does it sound like a generic arts organisation?
22When you submit a grant application written with AI assistance, are you competing fairly against other organisations that wrote theirs without that help?
23Does your AI-assisted application articulate your specific artistic vision, or does it say what you think funders want to hear?
24Have you compared a funding application you wrote yourself against one you drafted with ChatGPT to see where they actually differ?
25If a funder receives fifty applications and half were written with AI assistance, what happens to their ability to distinguish between organisations with real vision and those without?
26When AI polishes your application language, does it remove evidence of the real constraints or challenges your organisation faces?
27Are you using AI to make weak ideas sound better, or to express strong ideas more clearly?
28Does your grant writing describe what you actually do, or what you think will get funded?
29If you won funding partly because of an AI-optimised application, will you be able to deliver on what that application promised?
30What parts of your artistic practice or mission are hardest to explain to funders, and are those exactly the parts you should not let AI rewrite?
Institutional Culture and Values
31When your organisation adopts an AI tool for routine decisions, what human skill or judgement becomes harder to develop in your staff?
32Are junior staff learning to make curatorial or administrative judgements, or are they learning to follow what the AI recommends?
33If an AI system makes a decision that your organisation disagrees with, do you have the expertise left in-house to override it?
34What decisions in your organisation should never be delegated to an algorithm, even if the algorithm is more efficient?
35When you choose an AI tool, are you thinking about what your organisation will look like in ten years if you keep using it?
36Does your organisation have a genuine artistic perspective, or are you becoming what AI systems predict audiences will accept?
37What would it mean for your organisation to lose the ability to make a choice that no AI system would recommend?
38When you adopt AI tools, are you doing it because they solve a real problem, or because competitors are using them?
39If you stopped using AI systems tomorrow, could your organisation still function and make good decisions?
40What does cultural leadership mean if you are optimising for what algorithms predict rather than what you believe matters?
How to use these questions
Compare a decision you made with AI assistance against one you made without it. Look for the actual differences in quality, authenticity, and risk.
When an AI tool generates multiple options, pick the one that seems least obvious rather than the one that looks most polished. That is often where your real choice is.
Ask artists and curators in your network which AI tools they have stopped using and why. Those exit stories are more useful than adoption stories.
Before adopting any AI system for institutional decisions, write down what you would do if the system failed tomorrow. If you do not have a plan, you are too dependent.
Set a quarterly review where you examine one significant decision your organisation made and ask whether AI assistance made it better or just faster.