In September 2016, a short film called Sunspring premiered at the Sci-Fi London film festival. Its screenplay had been written entirely by an AI system called Benjamin — trained on dozens of science-fiction scripts. The actors read the lines straight. The film was strange, funny, and weirdly moving. It won an award.
The filmmaker, Oscar Sharp, didn't claim AI had replaced him. He said working with Benjamin felt like collaborating with "a very strange creative partner who had absorbed all of science fiction and had no idea what it meant."
Before we can understand what AI does to creativity, it helps to remember that creativity has never been purely "inside the head." Every artist uses tools. Painters use brushes and pigments. Writers use keyboards. Musicians use instruments — and for the last century, also recording equipment, synthesizers, and digital audio workstations.
Each new tool changed what was possible. The electric guitar didn't replace the musician — it gave musicians access to sounds that were literally impossible before. AI is best understood as the next major creative tool. What makes it different is that this tool can generate raw material: sentences, melodies, images, code. That's new. And it raises real questions worth thinking carefully about.
Researchers and practitioners have identified three broad ways humans and AI currently work together creatively. Understanding them helps you figure out where you fit.
Here's something important to understand: AI systems like large language models or image generators don't have ideas, feelings, or intentions. When an AI writes a poem, it is doing something statistically sophisticated — predicting which words, given a prompt, are most likely to produce something coherent and relevant, based on patterns it learned from enormous amounts of human text.
That sounds less magical when you say it that way. But consider: human creativity also involves pattern-matching. When a songwriter hears a chord progression and feels an emotion, that too involves the brain recognizing patterns from every song it has ever heard. The difference is that humans have lived experience, bodies, and meaning attached to those patterns. AI has patterns without experience. The creative meaning still comes from you.
AI doesn't create meaning. It generates material. The human decides whether that material means something — and that decision is the creative act.
In this lab you're going to have a real creative conversation with an AI assistant. Your goal is to explore one of the three creative models from Lesson 1 — Amplifier, Provocateur, or Co-Author — by actually trying it.
Start by telling the AI what kind of creative project you have in mind (writing, music, art, games — anything goes). Then work together to take it somewhere interesting. Try at least 3 exchanges.
In August 2022, Jason Allen entered a piece called Théâtre D'Opéra Spatial into the Colorado State Fair fine arts competition — and won the digital art category. He had used Midjourney, an AI image generator, but had spent weeks crafting, refining, and iterating on the prompts. He printed it on canvas.
The controversy that followed raised a real question: was the creative work in the final image, or in the hundreds of decisions Allen made in directing the AI toward that image? Most artists and critics who looked closely at his process concluded: the skill was in the prompting.
A prompt is essentially a set of instructions. But instructions can be vague or precise, boring or imaginative. The quality of what AI produces is directly tied to the quality of your prompt. This isn't a coincidence — it's how these systems work. They respond to the specifics of what you ask.
Think about the difference between asking a friend "make me something to eat" versus "can you make me a toasted sourdough sandwich with avocado, a poached egg, and a pinch of chilli flakes?" The second request gets you something much more interesting — and much closer to what you actually wanted.
Experienced AI creators have identified four elements that consistently improve outputs:
The winning Midjourney prompt Jason Allen used included terms like "operatic," "volumetric lighting," "intricate," "8K resolution," and referenced specific artistic styles. Each word was a deliberate creative decision — not a single click.
Professional AI artists almost never get their best result on the first try. The real creative work happens in the back-and-forth: trying a prompt, evaluating the output, identifying what's missing or wrong, and adjusting. This is iteration — and it's the same process professional designers, writers, and musicians have always used.
The difference is speed. With AI, you can go through twenty iterations in an afternoon that might have taken weeks with traditional tools. More iterations means more chances to stumble onto something genuinely unexpected and great.
Think of prompting like giving directions to a very literal-minded friend. If you say "go somewhere fun," they might end up at a dentist's office (they find it fun). But if you say "go to the skate park on Oak Street, skate the halfpipe, and try to land a 180," they know exactly where to go and what to do. Specificity is kindness — to both your friend and your AI.
In this lab, practice the four-element prompt framework from Lesson 2. Start with a weak, vague prompt. Then rebuild it with Subject, Style, Constraint, and Perspective. Ask the AI to evaluate your prompts and suggest improvements. Complete at least 3 exchanges.
The AI tutor in this lab is specifically focused on prompt crafting — it will ask you questions, suggest additions, and help you see why small changes make a big difference.
In April 2023, a track called Heart on My Sleeve appeared on Spotify and Apple Music. It featured voices that sounded unmistakably like Drake and The Weeknd. Neither artist had recorded a word of it. The creator, known as Ghostwriter977, had used AI voice-cloning tools to generate the vocals. The song racked up millions of streams before being removed. Universal Music Group demanded takedowns, and the Recording Academy ruled it ineligible for Grammy consideration.
The incident didn't settle the debate — it ignited it. Ghostwriter977 called it an experiment in the future of music. Major labels called it theft. Most listeners just thought it sounded great.
AI tools now exist across every stage of music production. Suno and Udio can generate complete songs — vocals, instrumentation, lyrics — from a text prompt. AIVA (Artificial Intelligence Virtual Artist), founded in 2016 and based in Luxembourg, composes original orchestral music and has had pieces performed by live orchestras. Holly Herndon, an American composer and musician, has built an AI vocal model of her own voice that other artists can license and use.
The Heart on My Sleeve case sits at the hard edge of this territory — the question of voice cloning without consent. But it also reveals something important: audiences couldn't tell the difference. That raises profound questions about authenticity, about what we're actually listening for when we love a song.
In March 2023, Amazon noticed a sudden surge in self-published books with AI-generated content. Within weeks, hundreds of titles attributed to fake authors appeared in its Kindle store, including a survival guide purportedly about a real wildfire that hadn't happened yet. Amazon had to introduce new AI-disclosure policies.
At the same time, many serious writers began using AI differently. Robin Sloan, author of the novel Mr. Penumbra's 24-Hour Bookstore, wrote publicly about using a custom-trained language model as a writing partner — not to generate prose, but to give him unexpected reactions to his own drafts. He described it as "the most interesting creative relationship I've had in years."
The contrast matters. AI used to flood a market with cheap content is a different thing from AI used as a thoughtful creative tool. The technology is the same. The intent and craft are not.
OpenAI released DALL-E in January 2021. Within two years, AI image generation had moved from a research curiosity to a tool used by film studios, advertising agencies, and independent artists worldwide. The animated film The Crow (2022 short by director Jessy Moussallem) used Midjourney for concept art. Nike ran campaigns using AI-generated imagery. Christie's auction house sold a portrait generated by the Paris-based collective Obvious for $432,500 in 2018 — one of the first major AI artwork sales.
For younger artists especially, these tools have dramatically lowered the barrier to visual storytelling. A student who can't yet draw photo-realistically can still produce high-quality visual concepts for a game, comic, or film project. The creative vision no longer requires years of technical skill to communicate visually.
If you've ever had a great idea for a game, a comic, or a movie but felt frustrated that you couldn't draw it or write it the way you imagined it — AI tools are starting to close that gap. Your vision matters more than your technical skill level right now. That's new. It's worth thinking about what you'd make if that barrier was removed.
Choose one creative field from Lesson 3 — music, writing, or visual art — and have a focused creative conversation about a project in that field. The AI in this lab knows about AI tools in each creative domain and can give you specific, practical guidance.
Ask about real AI tools you can use, get feedback on your ideas, or develop a concept together. Aim for at least 3 substantive exchanges.
In February 2023, the US Copyright Office issued a ruling on a comic book called Zarya of the Dawn, created by artist Kris Kashtanova. Kashtanova had used Midjourney to generate the images and written the story and text herself. The office's decision was nuanced: the text and arrangement of the book were protected by copyright, but the AI-generated images themselves were not — because copyright requires human authorship.
The ruling sent shockwaves through the creative AI community. It didn't ban AI art. It just said that the parts a human actually created could be protected, and the parts AI generated could not. The creative choice, not the technical execution, was what copyright law cared about.
Copyright law in most countries was built around the idea that a human being made something. The US Copyright Act requires "original works of authorship" — the courts have interpreted "authorship" to mean human authorship. This has created a genuine legal gap that AI creative work falls into.
The practical consequence: if you prompt an AI and use the output exactly as generated, in the United States you currently cannot copyright it. But if you significantly arrange, edit, select from, or combine AI outputs with your own creative choices, the resulting work may be protectable. Your creative decisions are what the law actually values.
In January 2023, a class-action lawsuit was filed against Stability AI, Midjourney, and DeviantArt by artists including Sarah Andersen, Kelly McKernan, and Karla Ortiz. The artists argued that these AI systems had been trained on billions of images scraped from the internet — including their own work — without permission or compensation. The case is still working through the courts.
This is a genuinely contested ethical area. Some argue that training on publicly available data is like a human artist learning from paintings in a museum. Others argue that directly imitating a specific artist's style using a model trained on their work is exploitation. There is no settled answer yet. But as a user of these tools, it's worth knowing where the images and text patterns come from.
LAION-5B, one of the major training datasets used by open image-generation models, contains approximately 5.85 billion image-text pairs scraped from the public internet. The images belonged to millions of individual photographers, illustrators, and artists who were not asked for permission.
Before AI, "original" in art typically meant that the work came from the unique perspective, experience, and skill of a specific person. It was traceable to a human source. AI complicates this without erasing it.
A useful way to think about it: AI-generated outputs are, in a sense, statistical averages of what humans have already made. They tend toward the expected, the competent, the "good enough." Genuinely original work — the kind that surprises people, that captures something true about a specific human experience — still requires a human perspective at its core. AI can help you get there faster. It can't feel what you've felt or see what you've seen.
For young creators especially, this is actually good news: your specific life, your specific view of the world, your specific voice — those are the things that AI cannot replicate and that make creative work matter.
You might be asked at school whether it's "cheating" to use AI for creative work. The honest answer is: it depends entirely on how you use it. Using AI to dump out a finished essay you claim as your own is not the same as using AI to brainstorm ideas, then writing the essay yourself. The first replaces your thinking. The second supports it. Know the difference — and be honest about which one you're doing.
In this lab, you'll engage with the ethical dimensions of AI creativity. Bring a real dilemma, a real opinion, or a real question — and work through it with an AI that will challenge your thinking, present different perspectives, and help you form a clearer view.
This is a discussion lab, not a "right answer" lab. The AI won't tell you what to think — it'll help you think better. Aim for at least 3 substantive exchanges.