AI can mimic Van Gogh. Can it replace the struggle that made him great?

“We as teachers have also got accustomed to AI vocabulary,” Ohri says. “We can easily sort out that this is not a student-written essay or project. They know how to bypass the detectors. They humanise their content,” she says, referring to lightly edited AI-generated text. “But they use software for it, too!”

What worries Ohri most is not dishonesty in the traditional sense but a learned dependence. “It is not just disrupting the education system. It’s completely making students slaves of AI. They are completely blinded by it,” she says.

This vanishing act, the erosion of the human signature, echoes far beyond the school walls, into the realms of art, literature, and the very marketplace of ideas. As artificial intelligence produces fluent prose and convincing images instantly, compressing what once took days of labour into seconds, we are forced to ask: What is lost when we remove effort from the equation of creation? And what must we fiercely protect? From an industry perspective, this frictionless efficiency is the point.

Gautam Mehra, founder of Winterline AI Research in Pune, which helps manufacturing firms identify and evaluate the right AI technology providers for real-world deployment, says, “Earlier , data scientists often spent days cleaning, preparing and analysing complex datasets. Today, AI tools compress much of that work into hours,” he says.

The paradox of effort

Poet WB Yeats once captured the central paradox of creative labour in Adam’s Curse: A line will take us hours maybe/ Yet if it does not seem a moment’s thought/ Our stitching and unstitching has been naught.

The value of the finished work, Yeats understood, is inextricably linked to the hidden effort that shaped it, the revision, the doubt, the false starts. The labour is meant to vanish into the seamlessness of the final product but its absence is a carefully crafted illusion. The effort is the substance. Artificial intelligence has shattered this balance. The “stitching and unstitching” Yeats described has been bypassed. The result arrives pre-smoothed, devoid of the cognitive friction that traditionally shaped both the work and the worker. Siddharth Kapila, author of Tripping Down The Ganga, acknowledges the seductive appeal of this friction. “It packages information beautifully,” he says of AI. But he warns of what lies beneath that polish: “Once you open the box, you’ll find a few shiny apples, a few mouldy ones too.”

The danger, says Kapila, lies in the presentation. AI-generated prose exudes a confidence that can dull our critical instincts. “You have to keep checking and rechecking,” he says. “The more you use it, the more you notice the errors.” He recounts asking a model for a simple factual list, like the last 10 chief ministers of Delhi.

“There will be repetitions, missing names and wrong ordering. Then it apologises and gives you another version.” Human research involves sifting through contradictory sources, weighing biases and building understanding slowly. AI collapses this essential process, offering a clean answer stripped of its messy, truthful context.

Mehra, however, frames AI tools as an enhancement of human capability, arguing that the tool’s utility outweighs its flaws. “Anything that is helping you reduce your mistakes is good,” he says. For him, the cognitive partnership is key: “AI doesn’t work by itself. It needs a human for commands.”

AI-generated image of Van Gogh (Photo: Freepik)

Atrophy of the cognitive pause

The erosion is most corrosive where thinking is still being formed. A 2025 study from MIT on essay writing found that participants who relied heavily on AI showed weaker recall and reduced engagement with their own arguments. They described feeling a detachment from their work.

Ohri sees this daily. “Students just blindly copy and submit the work,” she says. Kapila defines this as the loss of the “cognitive pause”— the mental grappling that occurs when faced with a blank page or a complex idea. “Even for something basic — an email or a short note — it does the thinking for you,” he says. “When you outsource that repeatedly, you’re not just saving time. You’re giving up a habit of thought.”

Writing has long been a primary tool for thinking. To write is to discover what you know, to wrestle chaos into coherence. When that process is delegated, thinking itself becomes superficial, not because of incapacity but from disuse. Kapila draws a crucial generational line. “It might not affect people who grew up writing without it,” he says. “But for those coming up now, who may not practise composing and revising, it becomes risky.” The articulation of thought, a foundational human skill, risks atrophy.

Mehra, however, interprets this not as a loss of intellect but as its new expression: “A person, if he is able to use his intellect to guide the AI to write an essay, book, or edit a photo, can create something wonderful.”

When tools dictate tone

The pressure towards smooth, frictionless prose is now embedded in the very tools we use. Generative AI has moved from novelty to infrastructure, sitting invisibly inside word processors, grammar checkers and editing software, offering assistance even when unsought. “I am anti-AI, especially in creative fields,” says Arunima Tenzin Tara, author of The Ex Daughters of Tolstoy House. “However, AI has crept into all our writing software. Even things people trusted now come with AI that is supposedly helping you frame sentences.”

This “help,” she says, is far from neutral. Suggestions to “improve clarity” or “rewrite in a more engaging tone” encode specific, often homogenising, assumptions about what “good” writing should be. Over time, these nudges exert a gentle, persistent pressure, sanding down idiosyncrasies and pushing writers toward a standardised, inoffensive, middle-of-the-road voice. The unique signature of the writer is eroded by an algorithm optimised for palatability.

This concerns editors like Pragya Mittal, who has worked as a copy editor in publishing houses across Delhi and has watched automated tools enter work flows. “AI gathers information and arranges it into something readable. But what it categorically lacks is judgement. Editing isn’t about cleaning a file; it’s about knowing when not to interfere”, Mittal argues.

Run a manuscript through aggressive automated tools and something vital is lost. The prose may become grammatically flawless but it also becomes less particular. Long, rhythmic sentences are chopped. Meaningful digressions are eliminated. “It reads too polished, too machine-like,” Mittal says. The editor’s role, to be a discerning human reader protecting the author’s voice, is undermined by a tool designed for uniform correction.

When volume drowns value

The logic of frictionless production, once unleashed, inevitably reshapes the economic landscape. When the cost of generating text or images approaches zero, volume can swiftly replace quality as a market driver. An industry focused on margins and algorithms does not necessarily need great books; it needs a constant, overwhelming stream of content.

Peruvian author Marie Arana experienced this first-hand after publishing her book LatinoLand. She soon found online marketplaces flooded with AI-generated “summaries” and “companion guides” bearing her book’s title. These were not direct plagiarism but parasitic imitations, synthetic texts designed to capture search traffic and monetise attention.

Mehra, too, draws a clear distinction between AI-assisted output and original human creation. “Originality will always be expensive,” he says. “A Van Gogh will always sell in millions. So you cannot take out that originality.” He believes the human intellect remains the non-replicable core: “You need to have intellectual capability. AI can just help you achieve things.”

Collaborator and potential usurper

The anxiety over the vanishing signature extends beyond the written word. In visual arts, questions of authorship, style and replacement are unfolding with striking parallels. Harshit Agarwal, an IIT graduate and artist who began working with AI at the MIT Media Lab, treats technology as both collaborator and potential usurper.

“Theoretically it is possible,” he says of AI replicating an artist’s style, “especially if you have a fixed pattern of working.” Many artists develop a recognisable visual language through repetition. “That is definitely something that an AI could replicate, particularly for artists with large, digitised bodies of work,” Agarwal notes.

His own practice consciously engages with this risk. He collects and curates thousands of images. For a project on gender in India, he trained AI on drawings made by people visualising their own understanding of gender, then trains small models to learn from them. “What is interesting artistically for me,” he says, “is the fact that collecting and curating the data gives me a lot of agency.” He sometimes pushes the AI with minimal information to produce glitchy, artefact-filled results, foregrounding the machine’s otherness rather than hiding it.

Yet the unease remains. Given enough data, AI could approximate not only an artist’s style but also mirror their patterns and habits of thought. The signature could be copied, leaving the original artist competing with a ghost of their own making.

The counter movement

Confronted with this rising tide of synthetic smoothness, a counter-movement is emerging, one that consciously reintroduces friction to preserve agency.

For Ohri, the strategy is a deliberate return to roots. While she uses AI to design interactive quizzes and classroom activities, she has instituted AI-free zones for core thinking. “I am again going to the traditional way of teaching,” she explains. “I give classroom work, stay with them, interact with them… I encourage debates and discussions where there will be no use of any technology. We just sit and talk about a given topic, then we brainstorm on it.”

In these unplugged spaces, something vital re-emerges. “Sometimes they come out with discussions that even I don’t have a pointer to,” Ohri says, her voice brightening. “That is something so unique and out of the box. But I am very sure that if I had given this topic as a work assignment, they would not have used their brain,” she adds. Her mission is to protect the environment where the cognitive pause can still occur, where a student’s own voice must be summoned, hesitant and alive.

AI cannot decide what is worth saying in the first place. It cannot originate a question born of lived experience. It cannot replicate the “living consciousness moving through the world” that Kapila sees as the core of true creative or academic work. It cannot replicate, as he says of his own journey in his book, Tripping Down the Ganga, “my contradictions… Why I took the journey, what I felt.” This fundamental limit is acknowledged even by proponents such as Mehra, who concludes that the tool’s power is contingent on human direction and intellect. Using the example of a complex creative work like Dan Brown’s The Da Vinci Code, he argues: “Will AI help him in writing it? No. He cannot, because a person’s intellectual power is not there… AI can help you if you know where to look.”

Mo Gawdat, former Chief Business Officer at Google, during a keynote address for South by Southwest (SXSW) Sydney, offered a model for a more critical, empowered engagement with AI. He argued that intelligence itself is neutral. “There is no inherent value of good or evil in intelligence,” he said. “It’s what you use it for.” His conditional optimism lies in using AI not as an oracle but as an interlocutor, running questions through multiple systems, comparing answers, probing for inconsistencies. “If you use AI that way,” Gawdat contends, “it’s the best thing that ever happened.” The key is using the tool to sharpen scepticism and not surrender to convenience.

Yeats understood that the value of a line of poetry lies in the hours it took, hours that vanish to leave only their transformative effect on the language. If that labour is eliminated entirely, not hidden but absent, something essential is lost with it. Our task in this new age is not to reject the tool, but to defend, with fierce intentionality, the irreplaceable human spaces where our signatures, in all their flawed and glorious individuality, are still forged.

The opposing view processes this as a necessary evolution. As Mehra puts it: “AI is the wave right now. If you’re not able to ride it, you will be left behind.” The tension between protecting the cognitive pause and embracing efficiency defines the fundamental struggle of the AI age.

With inputs from Trisha Mukherjee