Artificial intelligence will not replace creativity; it will exhibit our laziness

Pablo Picasso once dismissed computers in a sentence that has aged better than the machines he was rejecting. 'Computers,' he said in a 1968 interview with William Fifield, 'are useless. They can only give you answers.' Seven decades later, GPT-5 can write a sonnet to specification and BHASHINI can translate it into 22 Indian languages within a second. What no system has yet learnt is whether the sonnet should be written, for whom, and why. Artificial intelligence is the mirror in which the intellectual habits of a society — its depth or shallowness, its courage or its imitation — appear at scale. AI will not abolish human creativity. It will exhibit, painfully and publicly, what counts as creativity and what was always mere repetition. The mirror is unsparing, which is why it is useful.

The cognitive scientist Margaret Boden, in The Creative Mind (1990, revised 2004), distinguished three kinds of creativity. The first is combinational — pairing familiar ideas in unfamiliar ways. The second is exploratory — searching the edges of an existing conceptual space. The third is transformational — breaking the space itself and inventing a new one. Today's large language models excel at the first, are competent at the second, and remain weak at the third. They can re-arrange known forms with statistical sophistication; they cannot decide that the form itself needs changing. Original problem-formulation, contextual judgment under value-conflict, embodied empathy — these are the territories where humans still ask the questions before machines compute the answers. The poet who decides that a sonnet's volta should now appear in line nine rather than line eight is doing something that requires a lived sense of how meaning unfolds in time, a sense models can simulate statistically but not feel.

India's creative economy is being reshaped under these pressures. The ANI v OpenAI litigation, filed in the Delhi High Court in November 2024, became a landmark test of whether copyrighted news content can be ingested to train commercial language models. The case is pending; its outcome will guide the entire Indic-content licensing regime. The IndiaAI Mission, approved on 7 March 2024 with an allocation of 10,371 crore rupees over five years, aims to build sovereign compute, datasets and skilling. BHASHINI has become the world's most ambitious public Indic-language stack. The IT Intermediary Guidelines Amendment Rules of October 2025 require labelling of synthetic content. The Ministry of Electronics and Information Technology's 2025 consultation paper on a Hybrid Licence Model proposes a middle path between unrestricted training-data scrape and absolute creator royalty — a model that other developing economies are watching. AIRAWAT, the public AI supercomputer at C-DAC Pune, provides indigenous compute capacity that breaks dependence on hyperscaler-only training infrastructure.

Work itself is being reorganised. A World Bank study of 2024 estimated that around 5 per cent of jobs in India will be displaced by AI by 2030, while 12 per cent will be augmented. The implication is not catastrophic unemployment but a structural shift in the mix of human work toward the parts machines cannot do well: framing problems, defining values, building trust, choosing among ethically loaded options. Education must move from answer-recall to question-formation. The National Education Policy 2020 already emphasises critical thinking, conceptual understanding and inter-disciplinarity; its implementation through the Aspirational Skill India 4.0 mission will determine whether a generation of Indian students learns to use AI well or merely to use AI. The Atal Tinkering Labs in 10,000 schools, the National Initiative for Developing and Harnessing Innovations, and the proposed AI Centres of Excellence in agriculture, health and sustainability are all small pieces of the right infrastructure. The harder problem is teacher preparation — most Indian teachers themselves did not learn to formulate questions during their own schooling.

The counter-view is sharper than it appears. AI's pattern-recombination, sceptics argue, is creativity by another name. DeepMind's AlphaFold has predicted the structure of over 200 million proteins, a feat no human researcher could have matched. AlphaGo's 'move 37' in its 2016 match against Lee Sedol was a play no human grandmaster had considered. The 2023 Stable Diffusion image generators produce paintings that critics cannot distinguish from human work without metadata. If creativity is producing previously unseen valuable outputs, AI is creative. The line, the argument runs, is blurring.

The line is blurring, but it is not the line that matters most. AlphaFold's predictions are valuable because biologists asked the question 'how do proteins fold?' for sixty years. AlphaGo's brilliance is brilliance within a game humans invented. The human task — to ask which proteins matter for which disease, which game matters for which culture, which life matters at this moment — remains stubbornly ours. AI is becoming an extraordinary collaborator in answering chosen questions. The choosing remains a human act, and the laziness of leaving the choosing to the machine is exactly what AI exhibits when it is allowed to. A student who submits an AI-drafted essay she does not understand has not been replaced by AI; she has been exposed by it. A lawyer who files a brief with hallucinated citations is not careless because of AI; AI has only made his shortcut visible.

The Indian creative tradition has resources for this moment. The bhakti poets — Kabir, Mirabai, Tukaram — produced verse rooted in lived experience that no algorithm can simulate honestly because the algorithm has no life to begin with. Premchand's stories drew their power from a moral imagination shaped by particular villages and particular hungers. Satyajit Ray's cinema worked through the slow attention of a director who watched faces. These are not merely cultural products; they are demonstrations of what creativity has always been — perception, courage, and the willingness to be specific. AI can assist these processes; it cannot originate them, because origination requires being a particular person in a particular world.

R Buckminster Fuller's working principle was that 'you never change things by fighting reality'. The reality is that AI is here, and it exhibits us back to ourselves. The student who uses ChatGPT to write an essay she does not understand is not lazy because of AI; AI has only shown her habit. The lawyer who files a brief with hallucinated citations is not careless because of AI; AI has only made his shortcut visible. The republic that builds AI literacy — and AI humility — alongside AI capability will be a republic where creativity flourishes, not flees. The mirror is honest. The discipline is to look into it. India's quietly stated bet, across NEP 2020, IndiaAI Mission and BHASHINI, is that this republic will look — and re-learn, at scale, the difference between asking and answering. The civilisations that learn the difference will use AI to extend their creativity; those that do not will use AI to exhibit their decline.