Tom Holland said on Spain’s El Hormiguero this week that artists are “safe” from AI because creativity depends on human experience, emotions, and understanding, remarks reported June 17 by Variety and Deadline as Hollywood’s AI fight keeps widening. His line is comforting, and in one important sense it is true. But it is also the kind of truth that can distract from the less romantic problem: the machines do not need souls to change who gets paid.
Holland’s argument is a familiar one, especially among performers. AI can process data, he said, but it cannot understand the difference between happiness and sadness; art is not copying, it is expression. That is a clean moral distinction, and it lands because it describes what most people still want from film, music, games, and writing. The question for artists, though, is not whether AI has inner life. The question is whether studios, platforms, agencies, advertisers, and software vendors will treat “good enough” synthetic output as a substitute for human labor.
The actor’s comments arrived during promotional appearances with Zendaya, with Holland preparing for a busy summer that includes Christopher Nolan’s The Odyssey and Marvel’s Spider-Man: Brand New Day. That context matters. Holland is not a technologist issuing a policy paper; he is an actor defending the premise that performance and art come from lived experience.
In that frame, he is right. Generative AI systems do not possess grief, memory, embarrassment, loyalty, vanity, dread, or desire. They can simulate the statistical shapes of those things because they have been trained on oceans of human-made material, but they do not experience them. That distinction is not trivial; it is the core reason a human audience still searches a face, a voice, or a line reading for evidence of intention.
But the commercial entertainment business has never required every output to be art. It requires schedules, budgets, rights clearance, trailer variants, marketing copy, pitch decks, localization, rough concept passes, visual references, background assets, and endless internal versions that the audience never sees. This is where “AI has no soul” becomes a weaker defense than it first appears, because much of the industry’s work is not rewarded for soul in the first place.
The danger is not that AI becomes Tom Holland. The danger is that AI becomes the junior designer, the assistant editor, the background compositor, the draft copywriter, the previs vendor, the localization contractor, the temp researcher, and the first-pass storyboard artist. It is not an existential replacement of artistry so much as a steady downgrading of the labor ladder beneath it.
That is why actors, writers, directors, and production craftspeople have responded so differently to the same technology. Some see a tool that can accelerate previsualization, repair workflows, or enable smaller teams to execute ambitious ideas. Others see a laundering mechanism for copyrighted work and a management weapon for reducing headcount while preserving the appearance of creative abundance.
Both views can be true. A director using AI to sketch storyboards is not the same as a studio using an actor’s likeness without meaningful consent. A disabled performer using synthetic voice technology is not the same as a company scraping living artists’ work to build a commercial system that competes with them. The industry’s problem is that the same term, AI, now covers assistance, automation, imitation, compression, surveillance, and replacement.
Holland’s comments fit into a wider chorus. Guillermo del Toro has warned of image and cinema illiteracy. Some stars have urged colleagues to understand the technology rather than merely fear it. Martin Scorsese’s association with an AI company sparked criticism from art directors who saw endorsement where he saw a new tool for storyboarding. The common thread is not agreement; it is the dawning realization that Hollywood can no longer treat AI as a future-tense issue.
Markets do not always reward origin. They reward efficiency, leverage, novelty, and scale. A synthetic image does not need to be spiritually equivalent to a human painting to replace a mood board commission. A generated voice does not need to be a great performance to eliminate scratch recording work. A script-assist tool does not need to write Casablanca to reshape a writers’ room if executives believe it can generate drafts, variants, outlines, or notes fast enough to reduce staffing.
This is the part the WindowsForum audience will recognize from decades of enterprise IT. Automation rarely starts by replacing the most visible expert. It starts by hollowing out the repetitive, entry-level, or invisible work around the expert. The senior engineer keeps the title; the help desk shrinks. The principal architect still draws the blueprint; the junior documentation role disappears. The star actor remains the star; the ecosystem that trains future craftspeople becomes thinner.
That is why the “artists are safe” framing can become unintentionally elitist. Famous actors, prestige directors, and established writers may be protected by contracts, lawyers, fan followings, and brand value. The vulnerable layer is everyone whose creative labor is treated as replaceable input rather than authorship.
Microsoft’s AI strategy has made this unavoidable. Copilot-branded features increasingly present the PC as a place where documents, meetings, code, mail, images, and workflows can be summarized, drafted, searched, transformed, or automated through natural language. For some users, this is a genuine productivity gain. For others, it is an uncomfortable reminder that the software stack is being redesigned around a new assumption: the first draft, first summary, first image, first response, or first script may now come from a model.
Creative workers using Windows machines face the same split. AI features can clean audio, generate captions, retouch images, assist with code, produce layout options, and summarize research. Those tools can help a small creator compete with a larger shop. They can also help a larger shop demand more output from fewer people.
The Windows PC has always been a labor platform as much as a consumer device. Spreadsheets changed accounting departments. Desktop publishing changed print shops. Nonlinear editing changed post-production. IDEs and package managers changed software development. AI is another layer in that lineage, but it is different because it imitates the surface of judgment, taste, and language rather than merely accelerating calculation or layout.
Creative communities are not merely asking whether AI can feel. They are asking whether their work can be absorbed into systems that later compete with them. They are asking whether consent will be opt-in or assumed. They are asking whether “AI-assisted” credits will hide the scale of substitution. They are asking whether a young artist can still learn by doing the low-level work that models are now being sold to automate.
This is where the analogy to enterprise software becomes especially useful. No serious sysadmin would accept a black-box automation tool in a production environment simply because the vendor says it increases productivity. They would ask about logging, permissions, rollback, auditability, data retention, access controls, and failure modes. Hollywood needs the same mindset.
An AI-generated background plate is not just an image. It is a provenance problem. A synthetic voice is not just a sound file. It is a consent and identity problem. A model-generated script suggestion is not just a line of text. It is a labor, copyright, and credit problem. These are governance questions, not metaphysical ones.
But the phrase “AI is just a tool” is incomplete unless the next sentence explains who owns it, who trained it, who profits from it, and who can refuse it. A hammer does not ingest millions of carpenters’ work and produce a competing bid. A camera does not impersonate a living actor’s face without permission. A word processor does not quietly build a derivative market from every unpublished draft it touches.
The tool argument becomes honest only when it includes constraints. If AI helps a storyboard artist explore options faster, that is one thing. If it lets a producer skip hiring the storyboard artist at all, that is another. If a performer licenses a voice model for a specific project under a negotiated agreement, that is one thing. If a studio retains perpetual synthetic reuse rights buried in contract language, that is another.
For Windows users, the same principle applies at the desktop level. AI that runs with clear user intent, transparent data handling, and reversible changes is a productivity feature. AI that silently harvests context, blurs local and cloud processing, or makes compliance harder is an administrative risk.
That does not mean audiences do not care. It means their care is uneven. They may object fiercely when AI fakes a beloved actor, exploits a dead performer, or produces uncanny slop in a premium production. They may be indifferent when AI fills a background, drafts marketing copy, or powers an invisible recommendation system. They may even embrace AI when it allows personalization, accessibility, fan creation, or indie production that would otherwise be impossible.
This is why “AI cannot make real art” is not enough as a labor strategy. Some AI output does not need to be real art to compete for time. Attention is finite. If synthetic media fills feeds, search results, streaming thumbnails, social platforms, and game asset stores, it changes the market even when the best human work remains superior.
The future may not be a clean contest between human masterpieces and machine trash. It may be a swamp of mixed authorship: human-directed AI, AI-assisted human work, synthetic filler, licensed likenesses, unlicensed mimicry, and genuine craft marketed alongside automated content. The audience will need labels, but labels alone will not settle the labor question.
IT departments will be asked to deploy AI tools faster than policy can mature. Legal teams will worry about data leakage. Security teams will worry about prompt injection, sensitive documents, shadow AI accounts, and unclear retention. Managers will want dashboards showing time saved. Workers will wonder whether “time saved” becomes “headcount reduced.” Vendors will present all of this as inevitable.
The Hollywood debate strips away the euphemisms because the outputs are emotional. A synthetic actor feels more invasive than a synthetic spreadsheet summary. But the enterprise pattern is the same. AI turns accumulated human work into a reusable computational resource, and then the organization decides whether the benefit flows to workers, customers, shareholders, or vendors.
That is why admins should watch how creative unions handle consent and disclosure. A good AI policy for a studio and a good AI policy for a company intranet share some DNA. Both need permission boundaries. Both need audit trails. Both need clarity about what can be submitted to external systems. Both need a way to distinguish experimentation from production use.
The problem is that belief alone is not a policy. The creative industries need contracts, technical standards, provenance systems, licensing frameworks, and enforcement. They need to decide when AI is a brush, when it is a counterfeit press, and when it is an unlicensed archive wearing a friendly user interface.
Artists are not wrong to say AI lacks soul. They are wrong only if they assume the market will automatically reward soul over scale. History suggests otherwise. Human work survives technological shifts when people build institutions around it: unions, guilds, labels, certifications, platforms, audiences, and laws that make the distinction visible and valuable.
Holland’s argument is a familiar one, especially among performers. AI can process data, he said, but it cannot understand the difference between happiness and sadness; art is not copying, it is expression. That is a clean moral distinction, and it lands because it describes what most people still want from film, music, games, and writing. The question for artists, though, is not whether AI has inner life. The question is whether studios, platforms, agencies, advertisers, and software vendors will treat “good enough” synthetic output as a substitute for human labor.
Holland Draws the Right Line in the Wrong Place
The actor’s comments arrived during promotional appearances with Zendaya, with Holland preparing for a busy summer that includes Christopher Nolan’s The Odyssey and Marvel’s Spider-Man: Brand New Day. That context matters. Holland is not a technologist issuing a policy paper; he is an actor defending the premise that performance and art come from lived experience.In that frame, he is right. Generative AI systems do not possess grief, memory, embarrassment, loyalty, vanity, dread, or desire. They can simulate the statistical shapes of those things because they have been trained on oceans of human-made material, but they do not experience them. That distinction is not trivial; it is the core reason a human audience still searches a face, a voice, or a line reading for evidence of intention.
But the commercial entertainment business has never required every output to be art. It requires schedules, budgets, rights clearance, trailer variants, marketing copy, pitch decks, localization, rough concept passes, visual references, background assets, and endless internal versions that the audience never sees. This is where “AI has no soul” becomes a weaker defense than it first appears, because much of the industry’s work is not rewarded for soul in the first place.
The danger is not that AI becomes Tom Holland. The danger is that AI becomes the junior designer, the assistant editor, the background compositor, the draft copywriter, the previs vendor, the localization contractor, the temp researcher, and the first-pass storyboard artist. It is not an existential replacement of artistry so much as a steady downgrading of the labor ladder beneath it.
Hollywood’s AI Debate Has Moved Past Science Fiction
The rhetoric around AI in Hollywood still borrows too heavily from science fiction. The machine is imagined either as a soulless imposter or as a miraculous collaborator. In reality, the battle is already more bureaucratic: contract language, likeness rights, training data, residuals, model review periods, consent, and who gets to say no.That is why actors, writers, directors, and production craftspeople have responded so differently to the same technology. Some see a tool that can accelerate previsualization, repair workflows, or enable smaller teams to execute ambitious ideas. Others see a laundering mechanism for copyrighted work and a management weapon for reducing headcount while preserving the appearance of creative abundance.
Both views can be true. A director using AI to sketch storyboards is not the same as a studio using an actor’s likeness without meaningful consent. A disabled performer using synthetic voice technology is not the same as a company scraping living artists’ work to build a commercial system that competes with them. The industry’s problem is that the same term, AI, now covers assistance, automation, imitation, compression, surveillance, and replacement.
Holland’s comments fit into a wider chorus. Guillermo del Toro has warned of image and cinema illiteracy. Some stars have urged colleagues to understand the technology rather than merely fear it. Martin Scorsese’s association with an AI company sparked criticism from art directors who saw endorsement where he saw a new tool for storyboarding. The common thread is not agreement; it is the dawning realization that Hollywood can no longer treat AI as a future-tense issue.
The “Soul” Argument Comforts Artists but Not Workers
The strongest version of Holland’s point is philosophical: creativity is an act of human expression, and AI cannot originate meaning in the way humans do. The weakest version is economic: because AI cannot originate meaning, artists’ jobs are safe. The second claim does not follow from the first.Markets do not always reward origin. They reward efficiency, leverage, novelty, and scale. A synthetic image does not need to be spiritually equivalent to a human painting to replace a mood board commission. A generated voice does not need to be a great performance to eliminate scratch recording work. A script-assist tool does not need to write Casablanca to reshape a writers’ room if executives believe it can generate drafts, variants, outlines, or notes fast enough to reduce staffing.
This is the part the WindowsForum audience will recognize from decades of enterprise IT. Automation rarely starts by replacing the most visible expert. It starts by hollowing out the repetitive, entry-level, or invisible work around the expert. The senior engineer keeps the title; the help desk shrinks. The principal architect still draws the blueprint; the junior documentation role disappears. The star actor remains the star; the ecosystem that trains future craftspeople becomes thinner.
That is why the “artists are safe” framing can become unintentionally elitist. Famous actors, prestige directors, and established writers may be protected by contracts, lawyers, fan followings, and brand value. The vulnerable layer is everyone whose creative labor is treated as replaceable input rather than authorship.
Windows Users Are Already Living the Same Argument
For Windows users and IT pros, Hollywood’s AI fight is not some separate entertainment-world drama. It is the same debate now embedded in the operating system, productivity software, creative suites, developer tools, search, and collaboration platforms. The question is not whether AI has a soul in Word, Teams, Visual Studio, Photoshop, Premiere, or a browser sidebar. The question is whether it changes the value of the human sitting at the machine.Microsoft’s AI strategy has made this unavoidable. Copilot-branded features increasingly present the PC as a place where documents, meetings, code, mail, images, and workflows can be summarized, drafted, searched, transformed, or automated through natural language. For some users, this is a genuine productivity gain. For others, it is an uncomfortable reminder that the software stack is being redesigned around a new assumption: the first draft, first summary, first image, first response, or first script may now come from a model.
Creative workers using Windows machines face the same split. AI features can clean audio, generate captions, retouch images, assist with code, produce layout options, and summarize research. Those tools can help a small creator compete with a larger shop. They can also help a larger shop demand more output from fewer people.
The Windows PC has always been a labor platform as much as a consumer device. Spreadsheets changed accounting departments. Desktop publishing changed print shops. Nonlinear editing changed post-production. IDEs and package managers changed software development. AI is another layer in that lineage, but it is different because it imitates the surface of judgment, taste, and language rather than merely accelerating calculation or layout.
The Real Dispute Is Control, Not Consciousness
The soul question is emotionally powerful because it lets humans retain moral dignity. But in contract negotiations and production budgets, the more important word is control. Who controls the training data? Who controls a performer’s likeness? Who controls synthetic voice rights? Who controls disclosure? Who controls whether AI-generated material can enter a workflow without the knowledge of collaborators?Creative communities are not merely asking whether AI can feel. They are asking whether their work can be absorbed into systems that later compete with them. They are asking whether consent will be opt-in or assumed. They are asking whether “AI-assisted” credits will hide the scale of substitution. They are asking whether a young artist can still learn by doing the low-level work that models are now being sold to automate.
This is where the analogy to enterprise software becomes especially useful. No serious sysadmin would accept a black-box automation tool in a production environment simply because the vendor says it increases productivity. They would ask about logging, permissions, rollback, auditability, data retention, access controls, and failure modes. Hollywood needs the same mindset.
An AI-generated background plate is not just an image. It is a provenance problem. A synthetic voice is not just a sound file. It is a consent and identity problem. A model-generated script suggestion is not just a line of text. It is a labor, copyright, and credit problem. These are governance questions, not metaphysical ones.
The Tool Argument Needs Guardrails to Mean Anything
There is a reasonable pro-AI case in creative work, and it should not be dismissed. Artists have always used tools that once looked destabilizing. Photography changed painting. Synthesizers changed music. Digital editing changed film. Motion capture changed performance. Game engines changed previs and virtual production. The history of media is not a story of pure human hands untouched by machines.But the phrase “AI is just a tool” is incomplete unless the next sentence explains who owns it, who trained it, who profits from it, and who can refuse it. A hammer does not ingest millions of carpenters’ work and produce a competing bid. A camera does not impersonate a living actor’s face without permission. A word processor does not quietly build a derivative market from every unpublished draft it touches.
The tool argument becomes honest only when it includes constraints. If AI helps a storyboard artist explore options faster, that is one thing. If it lets a producer skip hiring the storyboard artist at all, that is another. If a performer licenses a voice model for a specific project under a negotiated agreement, that is one thing. If a studio retains perpetual synthetic reuse rights buried in contract language, that is another.
For Windows users, the same principle applies at the desktop level. AI that runs with clear user intent, transparent data handling, and reversible changes is a productivity feature. AI that silently harvests context, blurs local and cloud processing, or makes compliance harder is an administrative risk.
The Public May Not Care Until It Feels Cheated
One uncomfortable possibility for artists is that audiences will not consistently reject AI-generated media on principle. People say they value authenticity, but they also reward speed, novelty, memeability, price, and convenience. A synthetic trailer, parody, short film, or fan edit can command attention even if viewers know it is machine-generated.That does not mean audiences do not care. It means their care is uneven. They may object fiercely when AI fakes a beloved actor, exploits a dead performer, or produces uncanny slop in a premium production. They may be indifferent when AI fills a background, drafts marketing copy, or powers an invisible recommendation system. They may even embrace AI when it allows personalization, accessibility, fan creation, or indie production that would otherwise be impossible.
This is why “AI cannot make real art” is not enough as a labor strategy. Some AI output does not need to be real art to compete for time. Attention is finite. If synthetic media fills feeds, search results, streaming thumbnails, social platforms, and game asset stores, it changes the market even when the best human work remains superior.
The future may not be a clean contest between human masterpieces and machine trash. It may be a swamp of mixed authorship: human-directed AI, AI-assisted human work, synthetic filler, licensed likenesses, unlicensed mimicry, and genuine craft marketed alongside automated content. The audience will need labels, but labels alone will not settle the labor question.
IT Departments Should Read Hollywood’s Panic as a Preview
The reason this story belongs on a Windows community site is not that Tom Holland uses a PC or that Hollywood’s drama is inherently technical. It is because entertainment is giving us a visible version of a conflict that every knowledge-work industry is about to negotiate. When software can produce plausible drafts of the things people are paid to make, the first-order question is productivity. The second-order question is power.IT departments will be asked to deploy AI tools faster than policy can mature. Legal teams will worry about data leakage. Security teams will worry about prompt injection, sensitive documents, shadow AI accounts, and unclear retention. Managers will want dashboards showing time saved. Workers will wonder whether “time saved” becomes “headcount reduced.” Vendors will present all of this as inevitable.
The Hollywood debate strips away the euphemisms because the outputs are emotional. A synthetic actor feels more invasive than a synthetic spreadsheet summary. But the enterprise pattern is the same. AI turns accumulated human work into a reusable computational resource, and then the organization decides whether the benefit flows to workers, customers, shareholders, or vendors.
That is why admins should watch how creative unions handle consent and disclosure. A good AI policy for a studio and a good AI policy for a company intranet share some DNA. Both need permission boundaries. Both need audit trails. Both need clarity about what can be submitted to external systems. Both need a way to distinguish experimentation from production use.
Holland’s Optimism Still Has Value
It would be easy to dismiss Holland’s comments as celebrity comfort talk, but that would be unfair. His instinct protects something important: the belief that human experience remains the source of meaning. In a media environment flooded with optimization, automation, and franchise calculus, that belief is not naive. It is necessary.The problem is that belief alone is not a policy. The creative industries need contracts, technical standards, provenance systems, licensing frameworks, and enforcement. They need to decide when AI is a brush, when it is a counterfeit press, and when it is an unlicensed archive wearing a friendly user interface.
Artists are not wrong to say AI lacks soul. They are wrong only if they assume the market will automatically reward soul over scale. History suggests otherwise. Human work survives technological shifts when people build institutions around it: unions, guilds, labels, certifications, platforms, audiences, and laws that make the distinction visible and valuable.
The Spider-Man Star’s AI Lesson for the PC Era
Holland’s remarks are best read neither as a final answer nor as a mistake, but as a useful starting point. The practical takeaways are sharper than the headline debate about whether a machine can feel.- AI does not need consciousness to disrupt creative labor, because many jobs are vulnerable to automation long before the final product becomes fully synthetic.
- Famous performers may be better protected than the junior, freelance, and below-the-line workers whose tasks are easier to fragment and replace.
- Consent, disclosure, provenance, and auditability matter more than abstract arguments about whether AI output counts as art.
- Windows users and IT administrators should treat creative AI like any other enterprise automation layer, with clear rules for data, permissions, logging, and accountability.
- The best use of AI will augment human intent, but the worst use will quietly convert human culture into a cost-cutting substrate.
References
- Primary source: Variety
Published: Wed, 17 Jun 2026 23:39:00 GMT
Tom Holland Says Artists Are 'Safe' From AI
Tom Holland says artists are safe from AI taking their jobs because 'creativity has to do with the human experience.'
variety.com
- Independent coverage: Deadline
Published: Wed, 17 Jun 2026 21:22:00 GMT
Tom Holland Thinks "Creativity Is Safe From AI"
As the future of generative AI continues to loom over Hollywood, Tom Holland isn't worried about it coming for his job.
deadline.com
