MAGAZINE Voices Anand Neelakantan Ravi Shankar Neha Sinha Shampa Dhar-Kamath Dr Deepali Bhardwaj Swami Sukhabodhananda Buffet People Wellness Books Food Art & Culture Entertainment NEW DELHI september 21 2025 SUNDAY PAGES 12 Thinking Inside the Bot log in log out Fast Information Access Quick answers on many topics Not Always Accurate May provide outdated or incorrect info 24/7 Availability Always online and responsive No Real Understanding Doesn’t truly “know” or “think” Idea Generation Helpful for brainstorming and creativity Lacks Emotional Depth Simulates empathy, but doesn’t feel ChatGPT is everywhere. But is it a blessing or a curse? Learning Aid Explains complex topics in simple terms Productivity Tool Assists with writing, coding, summarising, etc Privacy Concerns Input may be stored or reviewed Dependency Risk Can reduce critical thinking if overused No Judgment Safe space to ask anything without embarrassment Limited Real-Time Data Needs plugins or tools for live info Lies Makes up gaps with imaginary content India now accounts for around 13.5% of all global ChatGPT usage, overtaking the US as the single largest user base India gets Chatting AI generated image Surveys indicate that about 36% of Indian users interact with ChatGPT daily, more than double the global average of 17% W By Tanisha Saxena hat do Arjun Mehta, caught in a relationship dilemma, and Priya Sharma, a student at Delhi University have in common? Both have the “bot”. They , turned to the same source for help—Chat Generative Pre-trained Transformer, or ChatGPT. For Arjun, who admits he was “spiralling like a Shakespearean hero” over a girl he was dating, ChatGPT didn’t offer a cheesy pickup line or a cliché like “just move on.” Instead, it served up a disarming question: “What’s your goal—connection, or reassurance?” That pause gave him perspective. On the other hand, Priya went down the slippery slope. During the mid-semester exams, she found herself overwhelmed. Desperate and short on time, she asked ChatGPT to write a 1,500-word essay for her assignment. A week later, her professor called her in. “This doesn’t sound like you,” he said. After further questioning and a review by the academic committee, she was suspended for the semester, her scholarship was withdrawn, and her academic record flagged. There was a time when advice lived in very human places. You cried on a friend’s shoulder, argued with a parent, scribbled down your worries for a therapist, or nervously dialed your doctor. These days, people are skipping all that and typing their most personal questions into a ChatGPT chat box. A bot becomes confidant, coach, sounding board, and sometimes even relationship referee. A new paper from OpenAI, Harvard, and Duke reveals that while ChatGPT now has 700 million weekly users sending 2.5 billion messages daily its usage has shifted from , primarily work-related tasks to more personal and informational queries. Also, by mid-2025, ChatGPT users shifted from being predominantly male to majority female, with nearly half of messages coming from users aged 18-25. Usage is growing four times faster in low-income countries than in high-income ones. The bot has taken over life. It has stealthily invaded private spaces of individuals into their deepest fears, insecurities and desires. It advises a desperate lover how to gain the attention of the object of his desire. It helps people with depression, or shyness, or despair with easy suggestions. It helps writers and designers to come up with quick plots and synopses for scripts to be submitted to publishers or OTT producers. It even enters the legal world to draft petitions and briefs. It does homework, writes PhD theses, and helps executives handle stress, or a nasty boss, or a difficult colleague. It drafts job application and recipes. So what is left? Total dependance on a bot that you’ve allowed to control your life. A bot that could take you down the better path or the garden path. It could end in promotions, marriage, or despair. Take interior designer Shuchi Jain, for instance. On a quiet evening in Indore, she leans over her desk. Dr Sandeep Vohra Rasshi Gurnani Founder, No Worry No Tension Healthcare Astrologer and psychologist “As a psychiatrist, I see one of the deepest concerns around ChatGPT being its subtle impact on how children and young adults learn to think.” “Today’s generation, especially those in their 20s and 30s, expects answers on demand. ChatGPT feels personal, almost like having a guide available anytime you need.” She is not sketching floor plans, but chatting—with a bot. She types in half-formed ideas for a client brief, and receives a cascade of neatly structured suggestions: mood board descriptions, material combinations, even references to design principles she hasn’t studied in years. For Jain, ChatGPT is an indispensable creative partner. “I put all the to-dos, tasks, events, and the site stage with process, which helps structure timelines, task lists, and project workflows, ensuring I don’t miss important steps,” she says. But are we starting to outsource too much? Ravi Iyer, a senior software engineer working in a tech firm in Bengaluru, learnt it the hard way . One evening, after returning from a short trip to Coorg, Ravi noticed a strange, circular rash on his thigh. Concerned, he asked ChatGPT to diagnose it based on the symptoms. The chatbot suggested it could be a mild fungal infection and recommended common antifungal creams. Over the next week, the rash spread rapidly When Ravi finally consulted . a dermatologist, the diagnosis stunned him: Lyme disease, likely contracted from a tick bite during his trip. The delay in treatment had allowed the infection to spread, requiring a longer and more aggressive course of antibiotics, and a two-week medical leave from work. In India—a nation where nearly half the population struggles with functional literacy and where access , to good teachers and healthcare remains scarce—the arrival of a free, conversational assistant has been nothing short of revolutionary Take . 17-year-old Renu from Baran, Rajasthan. “My English teacher is good, but she has 60 students in class. With ChatGPT, I can practise English whenever I want. It even explains grammar in Hindi if I get stuck,” she says. For 42-year-old shopkeeper Pramod Yadav in Gorakhpur, it is about dignity “Earlier I had to call . my nephew in Delhi every time I had to write an email for GST or order stock. Now I just ask ChatGPT. It writes everything perfectly he says. ,” In low-literacy contexts, many hesitate to approach teachers, bank officers, or government staff, worried about sounding foolish. ChatGPT offers anonymity People can ask it . anything and receive answers in seconds. But the same trust that empowers can also mislead. Experts worry that without a culture of cross-checking, misinformation can spread quickly “People often believe . what they see written, especially if it is in English,” says Dr Anjali Mehta, a sociologist studying technology use in rural India. There’s also the question of dependency A college . student in Lucknow admits that he uses ChatGPT for all his assignments. “Why waste time writing when it can do it for me?” The Digital Confidant Across India, where the stigma around relationship counselling still lingers in many social circles, an increasing number of urban millennials and Gen Zs are turning to ChatGPT for guidance in matters of the heart. From decoding mixed signals on dating apps to navigating arranged marriage setups, it is fast becoming a secret therapist, coach, and diary rolled into one. There’s no fear of judgment, no emotional burdening of another person, and no risk of gossip in tight-knit social groups where privacy is hard to come by . Yet, this trend also opens up questions. How reliable is a chatbot as a relationship counsellor? Is it ethical for people to lean on some Turn to page 2 I ndia is actively developing its own LLMs and chatbot systems tailored to its linguistic diversity, cultural contexts, and specific needs. But no Indian LLM has yet reached the global scale or sophistication of models like OpenAI’s ChatGPT. Sarvam AI, an Indian startup, is building foundational language models with a focus on Indian languages. Similarly, AI4Bharat, a government-supported research initiative, is developing models, datasets, and tools across all 22 constitutionally recognised Indian languages. They’ve released models such as IndicBERT and IndicBART, which are designed specifically for tasks in Indian languages. Another project, Paramanu, is a family of efficient generative models that support languages like Hindi, Tamil, Bengali, Odia, Telugu, and more. Despite this progress, Indian models face significant challenges. One of the most pressing is the lack of high-quality digital data in many Indian languages and dialects. Languages with low online presence or fewer annotated resources are particularly hard to model effectively. Another major barrier is computational cost. Training and deploying models at the scale of GPT-4 requires massive infrastructure and funding. Indian startups and labs often don’t have access to the kind of compute resources that organisations like OpenAI, Google, or Meta do. There are also technical and UX limitations: while ChatGPT offers advanced features like image input, file handling, and coding capabilities, most Indian models are still text-only or in the early stages of expanding into multimodal capabilities.