The funniest part of the Canadian housing market is everyone still pretending this is a “young people budgeting problem” and not a completely cooked economic structure.
You can do everything right and still end up 31 years old living like a university student.
"Just work harder"
Note: Sharing this here as an essay against many Anti-AI talking points and arguments.
In this essay, I will argue that AI outputs can be art and that AI prompters can be artists. I will support this claim by showing that many Anti-AI arguments depend on mistaken assumptions about what art is, what tools do, and what artistic authorship requires. I will begin by explaining why art should not be defined only by manual labour. Next, I will argue that prompting can involve real artistic intention, judgment, and revision. Lastly, I will reply to several common Anti-AI objections, including “the AI made it”, “prompting is not skill”, “AI art has no soul”, “AI art is theft”, and “AI art harms artists”.
To begin, art is not defined by the physical difficulty of making an object. Art is not simply the act of moving a brush, pressing a shutter, sculpting clay, or striking piano keys. Those may be methods of making art, but they are not what makes the result art. A photograph can be art even though the camera performs the optical capture. A film can be art even though hundreds of people, machines, lenses, software systems, lighting tools, and editing suites contribute to the final image. A collage can be art even though it rearranges existing material. A DJ set can be art even though it uses previously recorded sounds. Therefore, if someone says AI output cannot be art because the user did not manually render every pixel, that argument also threatens photography, cinema, digital painting, sampling, collage, and much of modern design.
The better question is not “Was a tool involved?”. The better question is: Was there meaningful human intention, selection, direction, and judgment involved in producing or presenting the final work? If the answer is yes, then there is at least a plausible case that the output can be art. This does not mean every AI output is art. A random image generated from a lazy prompt may be no more artistic than an accidental phone photo of the floor. But it does mean AI involvement alone cannot disqualify something from being art.
Prompting is not merely typing a sentence and receiving magic. Sometimes it is that simple, just as sometimes photography is merely pointing a phone at one's lunch. But no one concludes from bad or casual photography that photography itself is not art. The same standard should apply to AI. A weak prompt followed by immediate posting may be lazy. A careful process involving concept, reference, framing, revision, selection, masking, inpainting, editing, colour correction, composition, and rejection of poor outputs is a different matter. In that case, the human is not a passive spectator. The human is functioning as director, curator, editor, designer, and sometimes co-composer.
Consider a simple example. Suppose two people use the same AI image model. One types “cool dragon” and accepts the first image. Another spends hours trying to create a specific image: an old dragon curled around a ruined observatory, not as a monster, but as the last guardian of a forgotten astronomical civilization. They adjust lighting, mood, scale, architectural style, facial expression, colour palette, camera angle, symbolic details, and emotional tone. They reject fifty versions because the dragon looks too aggressive, the observatory looks too generic, or the image fails to suggest melancholy. Eventually, they arrive at something close to their intended vision. It seems strange to say the second person contributed nothing artistically. They may not have painted the scales by hand, but they made a long series of artistic decisions about what the work should be.
One might object that the AI still “made” the image. This objection seems plausible at first, but it confuses execution with authorship. Tools often execute. Cameras capture light. Synthesizers generate sound. Photoshop applies transformations. A 3D renderer calculates shadows, reflections, textures, and perspective. The artist does not personally compute every photon in a Pixar frame, yet we do not say the renderer is the artist. We understand that the renderer is a tool within a larger intentional process. Likewise, the fact that an AI system performs much of the low-level generation does not prove that the human user has no artistic role. It only proves that the human is not making the work by traditional manual technique.
Another objection is that prompting requires no skill. But this argument treats the lowest-skill use of a tool as the essence of the tool. A person can take a thoughtless photograph. A person can make a lazy collage. A person can trace badly. A person can use a synthesizer preset with no musical understanding. This does not show that photography, collage, drawing, or electronic music are not artistic mediums. It only shows that a medium can be used badly. AI is the same. There is low-effort AI use, and there is high-effort AI use. The existence of the first does not negate the second.
Skill in AI art often appears in different places. It appears in concept formation, visual literacy, iterative judgment, prompt control, reference selection, model choice, editing, compositing, taste, and knowing when an output fails. It is less like carving marble and more like directing a difficult actor who sometimes misunderstands every instruction. The prompter must guide a system that is powerful but unreliable. The system may produce beauty by accident, but turning accident into a coherent finished work requires judgment. In this sense, AI art often resembles photography: One does not create the landscape, the light, or the clouds, but one can still create the artwork by choosing the frame, timing, exposure, composition, and final treatment.
The “AI art has no soul” objection is also weaker than it appears. Art does not get its meaning solely from the internal feelings of the tool. A paintbrush has no soul. A camera has no soul. A violin has no soul. A word processor has no soul. The relevant question is whether the human being using the tool can express intention, taste, humour, grief, beauty, anger, irony, or criticism through the resulting work. If a person uses AI to create an image about loneliness, memory, political absurdity, religious doubt, childhood fear, or cosmic wonder, the meaning comes from the human context and the human act of selection and presentation. The machine does not need a soul for the artwork to carry human meaning.
A related objection is that AI outputs are merely random. But this is also too simple. AI generation has stochastic elements, but randomness has long been part of art. Photographers rely on unpredictable light and street scenes. Painters use happy accidents. Musicians improvise. Surrealists used automatic methods. Writers discover meanings they did not consciously plan. Randomness does not eliminate artistry when the artist selects, shapes, interprets, and finalizes the result. The marble has veins. The watercolour bleeds. The camera catches an unexpected expression. The AI generates variations. In all these cases, the artist’s role is not to control every atom, but to recognize and shape what matters.
Another Anti-AI objection is not really about whether AI outputs can be art. It is about training data, consent, compensation, and labour disruption. These are serious issues. However, even if one believes some current AI training practices are ethically flawed, it does not follow that AI outputs cannot be art or that AI users cannot be artists. Those are separate questions. A painting made with a brush from an unethical factory can still be a painting. A film made on exploitative equipment can still be a film. A song made with uncleared samples can still be music, even if there may be legal or ethical problems surrounding it. The ethics of production matter, but they do not automatically settle the ontology of the artwork.
The claim “AI art is theft” also needs more precision. If an AI output directly copies a specific image, substantially reproduces a protected work, or is marketed deceptively as another artist’s labour, then the criticism is much stronger. But if a user creates an image that does not meaningfully reproduce a specific protected work and does not pretend to be someone else’s work, then calling it theft becomes far less obvious. Human artists learn from thousands of images, genres, influences, and conventions. This does not mean machine learning and human learning are morally identical. They are not. But it does mean the argument must be made carefully. “Influence”, “style”, “training”, “copying”, “market substitution”, and “plagiarism” are not the same concept.
Another objection is that AI art harms working artists by flooding markets with cheap images. This is a real concern, but it is not an argument that AI art is not art. Photography harmed some portrait painters. Recorded music changed live performance. Desktop publishing changed graphic design. Digital cameras changed photography. Streaming changed music and film. These disruptions can be painful and unfair, but they do not prove the new medium is not artistic. They prove that society has to deal with the economic consequences of new tools. The solution may involve disclosure norms, labour protections, licensing systems, new business models, and ethical standards. It should not involve pretending that no meaningful artistic activity is happening.
One might also say that AI prompters are more like commissioners than artists. Sometimes this is true. If someone merely says “make me a logo” and accepts the result, they are closer to a client than an artist. But many cases are not like this. The distinction depends on the degree of creative control and artistic decision-making. A film director does not personally act every role, build every set, sew every costume, compose every note, and render every visual effect. Yet the director can still be an artist because the director shapes the whole. Likewise, an AI user who develops the concept, directs the process, rejects failed attempts, modifies the output, and decides the final form can plausibly be an artist. The relevant category is not “manual labourer”. It is “creative agent”.
This is where many Anti-AI arguments become too blunt. They want one rule: “If AI was used, the human did not make it”. But this ignores the continuum of involvement. There is a major difference between typing one vague prompt and posting the first result, writing a detailed prompt and selecting the best of several outputs, iterating dozens of times toward a specific artistic intention, combining AI output with drawing or editing, and using AI as one stage in a larger creative workflow.
It is unreasonable to treat all these cases as identical. The more intention, selection, revision, and integration are involved, the stronger the case for human artistry becomes. This does not mean the AI user deserves credit for skills they did not exercise. An AI prompter should not pretend to have hand-painted an image if they did not. But it also does not follow that they deserve no credit at all. Credit should track actual contribution. If the contribution is concept, direction, iteration, editing, and final selection, then that is the contribution that should be acknowledged.
The mistake is thinking that art requires total control. It does not. No artist has total control. The painter must deal with the behaviour of paint. The photographer must deal with light, sensor, lens, subject, and timing. The actor must deal with body, voice, memory, and audience. The writer must deal with language, genre, convention, and unconscious association. Art is almost always a negotiation between intention and resistance. AI is simply a new kind of resistance: A strange, statistical, semi-obedient collaborator that often fails in interesting ways. The artist’s task is to direct that resistance toward meaning.
Therefore, the most reasonable conclusion is not that all AI outputs are art, nor that all AI users are artists. That would be too broad. The better conclusion is that some AI outputs are art, and some AI prompters are artists, depending on the role of human intention, judgment, revision, and presentation. This is the same kind of distinction we already make with photography, film, music production, collage, design, and digital art.
The Anti-AI side is right to worry about exploitation, deception, spam, market flooding, and disrespect toward human artists. Those concerns should be taken seriously. But they do not justify the stronger claim that AI outputs cannot be art or that AI prompters cannot be artists. That stronger claim rests on a narrow and historically fragile idea of art as manual execution. Once we recognize that art can involve direction, selection, framing, transformation, and meaning-making through tools, the categorical rejection of AI art collapses.
AI does not erase human artistry. It changes where some of the artistry happens.
In this essay, I will argue that AI outputs can be art and that AI prompters can be artists. I will support this claim by showing that many Anti-AI arguments depend on mistaken assumptions about what art is, what tools do, and what artistic authorship requires. I will begin by explaining why art should not be defined only by manual labour. Next, I will argue that prompting can involve real artistic intention, judgment, and revision. Lastly, I will reply to several common Anti-AI objections, including “the AI made it”, “prompting is not skill”, “AI art has no soul”, “AI art is theft”, and “AI art harms artists”.
To begin, art is not defined by the physical difficulty of making an object. Art is not simply the act of moving a brush, pressing a shutter, sculpting clay, or striking piano keys. Those may be methods of making art, but they are not what makes the result art. A photograph can be art even though the camera performs the optical capture. A film can be art even though hundreds of people, machines, lenses, software systems, lighting tools, and editing suites contribute to the final image. A collage can be art even though it rearranges existing material. A DJ set can be art even though it uses previously recorded sounds. Therefore, if someone says AI output cannot be art because the user did not manually render every pixel, that argument also threatens photography, cinema, digital painting, sampling, collage, and much of modern design.
The better question is not “Was a tool involved?”. The better question is: Was there meaningful human intention, selection, direction, and judgment involved in producing or presenting the final work? If the answer is yes, then there is at least a plausible case that the output can be art. This does not mean every AI output is art. A random image generated from a lazy prompt may be no more artistic than an accidental phone photo of the floor. But it does mean AI involvement alone cannot disqualify something from being art.
Prompting is not merely typing a sentence and receiving magic. Sometimes it is that simple, just as sometimes photography is merely pointing a phone at one's lunch. But no one concludes from bad or casual photography that photography itself is not art. The same standard should apply to AI. A weak prompt followed by immediate posting may be lazy. A careful process involving concept, reference, framing, revision, selection, masking, inpainting, editing, colour correction, composition, and rejection of poor outputs is a different matter. In that case, the human is not a passive spectator. The human is functioning as director, curator, editor, designer, and sometimes co-composer.
Consider a simple example. Suppose two people use the same AI image model. One types “cool dragon” and accepts the first image. Another spends hours trying to create a specific image: an old dragon curled around a ruined observatory, not as a monster, but as the last guardian of a forgotten astronomical civilization. They adjust lighting, mood, scale, architectural style, facial expression, colour palette, camera angle, symbolic details, and emotional tone. They reject fifty versions because the dragon looks too aggressive, the observatory looks too generic, or the image fails to suggest melancholy. Eventually, they arrive at something close to their intended vision. It seems strange to say the second person contributed nothing artistically. They may not have painted the scales by hand, but they made a long series of artistic decisions about what the work should be.
One might object that the AI still “made” the image. This objection seems plausible at first, but it confuses execution with authorship. Tools often execute. Cameras capture light. Synthesizers generate sound. Photoshop applies transformations. A 3D renderer calculates shadows, reflections, textures, and perspective. The artist does not personally compute every photon in a Pixar frame, yet we do not say the renderer is the artist. We understand that the renderer is a tool within a larger intentional process. Likewise, the fact that an AI system performs much of the low-level generation does not prove that the human user has no artistic role. It only proves that the human is not making the work by traditional manual technique.
Another objection is that prompting requires no skill. But this argument treats the lowest-skill use of a tool as the essence of the tool. A person can take a thoughtless photograph. A person can make a lazy collage. A person can trace badly. A person can use a synthesizer preset with no musical understanding. This does not show that photography, collage, drawing, or electronic music are not artistic mediums. It only shows that a medium can be used badly. AI is the same. There is low-effort AI use, and there is high-effort AI use. The existence of the first does not negate the second.
Skill in AI art often appears in different places. It appears in concept formation, visual literacy, iterative judgment, prompt control, reference selection, model choice, editing, compositing, taste, and knowing when an output fails. It is less like carving marble and more like directing a difficult actor who sometimes misunderstands every instruction. The prompter must guide a system that is powerful but unreliable. The system may produce beauty by accident, but turning accident into a coherent finished work requires judgment. In this sense, AI art often resembles photography: One does not create the landscape, the light, or the clouds, but one can still create the artwork by choosing the frame, timing, exposure, composition, and final treatment.
The “AI art has no soul” objection is also weaker than it appears. Art does not get its meaning solely from the internal feelings of the tool. A paintbrush has no soul. A camera has no soul. A violin has no soul. A word processor has no soul. The relevant question is whether the human being using the tool can express intention, taste, humour, grief, beauty, anger, irony, or criticism through the resulting work. If a person uses AI to create an image about loneliness, memory, political absurdity, religious doubt, childhood fear, or cosmic wonder, the meaning comes from the human context and the human act of selection and presentation. The machine does not need a soul for the artwork to carry human meaning.
A related objection is that AI outputs are merely random. But this is also too simple. AI generation has stochastic elements, but randomness has long been part of art. Photographers rely on unpredictable light and street scenes. Painters use happy accidents. Musicians improvise. Surrealists used automatic methods. Writers discover meanings they did not consciously plan. Randomness does not eliminate artistry when the artist selects, shapes, interprets, and finalizes the result. The marble has veins. The watercolour bleeds. The camera catches an unexpected expression. The AI generates variations. In all these cases, the artist’s role is not to control every atom, but to recognize and shape what matters.
Another Anti-AI objection is not really about whether AI outputs can be art. It is about training data, consent, compensation, and labour disruption. These are serious issues. However, even if one believes some current AI training practices are ethically flawed, it does not follow that AI outputs cannot be art or that AI users cannot be artists. Those are separate questions. A painting made with a brush from an unethical factory can still be a painting. A film made on exploitative equipment can still be a film. A song made with uncleared samples can still be music, even if there may be legal or ethical problems surrounding it. The ethics of production matter, but they do not automatically settle the ontology of the artwork.
The claim “AI art is theft” also needs more precision. If an AI output directly copies a specific image, substantially reproduces a protected work, or is marketed deceptively as another artist’s labour, then the criticism is much stronger. But if a user creates an image that does not meaningfully reproduce a specific protected work and does not pretend to be someone else’s work, then calling it theft becomes far less obvious. Human artists learn from thousands of images, genres, influences, and conventions. This does not mean machine learning and human learning are morally identical. They are not. But it does mean the argument must be made carefully. “Influence”, “style”, “training”, “copying”, “market substitution”, and “plagiarism” are not the same concept.
Another objection is that AI art harms working artists by flooding markets with cheap images. This is a real concern, but it is not an argument that AI art is not art. Photography harmed some portrait painters. Recorded music changed live performance. Desktop publishing changed graphic design. Digital cameras changed photography. Streaming changed music and film. These disruptions can be painful and unfair, but they do not prove the new medium is not artistic. They prove that society has to deal with the economic consequences of new tools. The solution may involve disclosure norms, labour protections, licensing systems, new business models, and ethical standards. It should not involve pretending that no meaningful artistic activity is happening.
One might also say that AI prompters are more like commissioners than artists. Sometimes this is true. If someone merely says “make me a logo” and accepts the result, they are closer to a client than an artist. But many cases are not like this. The distinction depends on the degree of creative control and artistic decision-making. A film director does not personally act every role, build every set, sew every costume, compose every note, and render every visual effect. Yet the director can still be an artist because the director shapes the whole. Likewise, an AI user who develops the concept, directs the process, rejects failed attempts, modifies the output, and decides the final form can plausibly be an artist. The relevant category is not “manual labourer”. It is “creative agent”.
This is where many Anti-AI arguments become too blunt. They want one rule: “If AI was used, the human did not make it”. But this ignores the continuum of involvement. There is a major difference between typing one vague prompt and posting the first result, writing a detailed prompt and selecting the best of several outputs, iterating dozens of times toward a specific artistic intention, combining AI output with drawing or editing, and using AI as one stage in a larger creative workflow.
It is unreasonable to treat all these cases as identical. The more intention, selection, revision, and integration are involved, the stronger the case for human artistry becomes. This does not mean the AI user deserves credit for skills they did not exercise. An AI prompter should not pretend to have hand-painted an image if they did not. But it also does not follow that they deserve no credit at all. Credit should track actual contribution. If the contribution is concept, direction, iteration, editing, and final selection, then that is the contribution that should be acknowledged.
The mistake is thinking that art requires total control. It does not. No artist has total control. The painter must deal with the behaviour of paint. The photographer must deal with light, sensor, lens, subject, and timing. The actor must deal with body, voice, memory, and audience. The writer must deal with language, genre, convention, and unconscious association. Art is almost always a negotiation between intention and resistance. AI is simply a new kind of resistance: A strange, statistical, semi-obedient collaborator that often fails in interesting ways. The artist’s task is to direct that resistance toward meaning.
Therefore, the most reasonable conclusion is not that all AI outputs are art, nor that all AI users are artists. That would be too broad. The better conclusion is that some AI outputs are art, and some AI prompters are artists, depending on the role of human intention, judgment, revision, and presentation. This is the same kind of distinction we already make with photography, film, music production, collage, design, and digital art.
The Anti-AI side is right to worry about exploitation, deception, spam, market flooding, and disrespect toward human artists. Those concerns should be taken seriously. But they do not justify the stronger claim that AI outputs cannot be art or that AI prompters cannot be artists. That stronger claim rests on a narrow and historically fragile idea of art as manual execution. Once we recognize that art can involve direction, selection, framing, transformation, and meaning-making through tools, the categorical rejection of AI art collapses.
AI does not erase human artistry. It changes where some of the artistry happens.