ChatGPT Comes to the Classroom

THAT OpenAI FOUNDER Sam Altman is an eccentric brainiac is an understatement. He is capable of referring to an AI “bubble” in the offing and at the same time confidently speak about the trillions of dollars in investment that he wishes to attract to construct new data centres as his company expands at a fast clip. He acknowledges the hype around artificial intelligence (AI) and still exudes optimism when it comes to his plans.

Recently, ahead of his visit to India this month, he showed the pragmatic side to his growth dream in what economists and analysts see as an effort by OpenAI—which has ballooned in size since the November 2022 launch of its flagship offering ChatGPT—to rise to new economic realities. ChatGPT is arguably the most talked-about conversational chatbot that uses large language models (LLMs) to generate human-like text and dialogue in response to user prompts.

The tech giant backed by Microsoft is setting up an office in New Delhi, hiring local talent, and has launched a ₹399 monthly plan to make ChatGPT more affordable for Indian users. There are also reports in a section of the media that OpenAI will set up a data centre in India—its second-largest market after the US—where it also has the largest base of student users, aligning itself with the government’s IndiaAI Mission worth $1.2 billion that envisages making home-grown LLMs to boost national AI capabilities. Although the company is tight-lipped about the location and other details, including the workforce, it is looking to hire in India. Bloomberg says that a new data centre could be significant for OpenAI’s Stargate-branded AI infrastructure push. Outside the US and besides India, OpenAI is looking for collaborations in Abu Dhabi, Norway and other countries. SoftBank, MGX, Oracle, and OpenAI itself are all involved in funding Stargate, and OpenAI is working with the US government to build an AI infrastructure to counter any threat from China.

The likes of Eli Etherton, the wellknown tech educator also known as ‘Eli the Computer Guy’, poke fun at the mere mention of “trillions of dollars” and financial instruments that can help meet such gigantic sums. Some others also state that OpenAI wants to do what Microsoft once did with Windows: quietly colonise young psyches.

Notwithstanding such pessimism from some quarters, Indian analysts and tech book authors who closely follow the AI growth trajectory say that OpenAI’s entry into India is shaped by new challenges. Bengaluru-based author and AI expert Sreejith Sreedharan says that the American startup bears high development and computing costs even as revenues grow quickly, creating an urgent need to reach new markets at scale. “India offers exactly that. With more than 825 million mobile broadband users, it is one of the most connected populations in the world, and mobile devices are the primary gateway for AI technology adoption.”

Sreedharan’s enthusiasm is shared by none other than Ashwini Vaishnaw, Union minister for electronics and information technology, who recently said that OpenAI’s decision to establish a presence in India reflects the country’s growing leadership in digital innovation and AI adoption. “With strong investments in digital public infrastructure, AI talent, and enterprise-scale solutions, India is uniquely positioned to drive the next wave of AI-led transformation. As part of the IndiaAI Mission, we are building the ecosystem for trusted and inclusive AI, and we welcome OpenAI’s partnership in advancing this vision to ensure the benefits of AI reach every citizen,” he said. The statement came in the wake of OpenAI’s launch of the following plans: ChatGPT Go, an India-specific subscription tier priced at ₹399 per month with integrated UPI payments; OpenAI Academy, a nationwide AI literacy initiative in partnership with the Ministry of Electronics and Information Technology; enhanced Indic language support; and the recently launched Study Mode feature in ChatGPT that the company says has gained significant traction in India.

Clearly, OpenAI’s immediate target is India’s education sector, which is plagued by endless odds, including elitism. The World Inequality Database, a research centre at the Paris School of Economics, notes that educational inequality is high in India. Its report states: “India has a larger share of tertiary-educated graduates while still grappling with a substantial illiteracy rate. India’s neglect of compulsory primary education left much of its population trapped in low-productivity agriculture.” The Indian education system also suffers from outdated curricula, poor infrastructure and funding, besides a huge gap between the skills learnt in schools and those required by industry. All these factors have, over the past decades, led to privileged students heading overseas for both education and jobs.

This sector, evidently, requires a lot of fixing.

It is in this context that Sreedharan says that the education sector in India adds another layer of exciting potential for OpenAI. “India has close to 1.5 million schools, over 265 million school students, and more than 45,000 colleges serving around 43 million higher-education learners. This staggering scale provides a direct route to bring AI into classrooms, equip teachers with new tools, and prepare students for a changing world of work. Partnerships with schools, universities, and skilling programmes could combine commercial value with significant social impact. AI in education pilots at schools championed by the World Bank in Africa have shown extraordinary promise, especially amongst students with meagre means.” He states that the Indian education market, both schools and colleges, mirrors challenges in Africa. “The promise and potential are quite alluring for transformative AI tech companies. OpenAI’s decision to introduce lower subscription pricing, such as the ₹399 per month plan, shows a clear adaptation to India’s price-sensitive market,” he avers.

It makes sense to see OpenAI’s India expansion plan as a necessity for the company and a win-win situation. Let’s not forget what OpenAI itself has said about its newer challenges. OpenAI CFO Sarah Friar said in a recent interview that the company faces “ongoing pressures due to the demand for computing power required for artificial intelligence”. In an interview with CNBC, she added: “It is voracious right now for GPUs and for compute…That’s why we launched Stargate. That’s why we’re doing the bigger builds.” The key lies here: according to Friar, the growing demand for computing power calls for more partners to diversify risk and increase supply.

Sreedharan notes that alongside education, India’s creator economy, already influencing hundreds of billions in consumer spending, represents another fast-growing segment where AI can accelerate productivity and creativity. “Even a modest foothold across students, educators, and creators can extend OpenAI’s growth path. With localised products, affordable options, and strong institutional links, India could become a central pillar of its global strategy,” he says.

This is why OpenAI is exploring local partners to set up a data centre in India, as reported by Bloomberg. The San Francisco-based company has also announced a major initiative to provide 5 lakh free ChatGPT licences for six months to students and teachers in India, covering government schools from classes 1 to 12, engineering and technical institutes, as well as K-12 (meaning primary to secondary education system) educators.

AS DOUBTS PERSIST about the use of generative AI in education, about it being high-tech plagiarism, and about legal hurdles that companies like OpenAI and others may face from entities that accuse them of stealing content and ideas, Open spoke to Subhashis Banerjee, professor and head of the Department of Computer Science, Ashoka University, to get a better understanding of the situation.

An alumnus of the Indian Institute of Science, Banerjee, whose areas of interest are computer vision and machine learning, points out that the “Gen AI tools” are great for learning, provided they are used judiciously. “A student can use it effectively to get a concept presented in different formats and styles for better understanding, generate examples and illustrations, juxtapose with alternate interpretations, and do critical analysis. They can also use these to search and summarise, test out ideas, and generate suggestive solutions for problems. Finally, Gen AI tools are great for getting routine tasks like coding, data format conversion, data presentation, etc done efficiently.”

Banerjee adds, “These tools are most often trained on internet data, where Indian data may be under-represented. Hence, they may not be adequately trained for the Indian context. A student will have to keep this caveat in mind. Also, Gen AI outputs are not always correct, and these systems are known to hallucinate sometimes. Consequently, these tools may turn out to be risky for the not-so-discerning students.”

For his part, Banerjee sees many advantages in educational institutions partnering with OpenAI. “Yes, this can be particularly helpful to make Gen AI offerings more representative by bringing in more local context.” His grouse is that as of now, an overwhelming majority of electronic text is in English written in the West, and, consequently, there is an obvious cultural bias. “Apart from Hindi, Bengali, Tamil, and possibly Marathi, the corpus of digitised text for the other Indian languages is limited. Consequently, the cultural contexts represented by most Indian languages will be underrepresented in the Gen AI tools,” he avers.

As regards Noam Chomsky’s criticism of ChatGPT and such offerings being high-tech plagiarism, Banerjee remarks, “I agree with Noam Chomsky’s thesis that ChatGPT-like systems are very far from Artificial General Intelligence [AGI] or self-awareness or cognition.” He adds that, however, it is undeniable that the semantic encodings in these Gen AI models are, to a large extent, language independent, enabling translation to new languages with a small corpus with very little training. “The same encodings also enable synthesising a concept in a new style or language, sometimes even as poems or creative art. Clearly, these encodings represent concepts at a much higher level than language syntax or grammar.”

He elaborates: “Chomsky believed that human language is primarily syntax-driven: there is an internal, generative system of rules [‘Universal Grammar’] that produces structured sentences, and meaning arises from mapping these structures into the conceptual system of the mind. Therefore, he believed that grammar is indispensable for linguistic meaning. ChatGPT-like systems do not have explicit grammar rules; they learn from surface statistical patterns of text [word co-occurrence, context, embeddings]. But they can still produce coherent, meaningful, and contextually appropriate sentences and translate concepts across languages and forms. Clearly, at the surface level, they seem to provide a counter-example for Chomsky’s thesis. On whether they disprove Chomsky entirely, I think the jury is still out.”

Meanwhile, Raghav Gupta, head of Education, India & Asia Pacific, OpenAI, posted on a social-media platform about the prospects of every student getting a 24/7 AI tutor thanks to OpenAI’s new initiatives. “[The] launch of the OpenAI Learning Accelerator was incredibly energising and showed us just how deeply this vision resonates. Parents and educators told us they don’t just want ‘answers’—they want AI as a true learning companion. And that’s exactly what we’re building with partners like IIT Madras, AICTE, and ARISE,” the post said. As of now, OpenAI has announced partnerships with the Indian Institute of Technology Madras (IIT Madras), All India Council for Technical Education, and so on. Professor V Kamakoti, director of IIT Madras, didn’t immediately respond to specific questions from Open about the goals of this partnership.

The comments from academia about OpenAI’s partnerships in education cut both ways.

Bindu Menon, associate professor of Media Studies, Azim Premji University, says, “The promise that OpenAI has offered is personalised learning like ‘study mode’ features and support to teaching material and address teacher shortage. However, this comes with challenges like digital inequality, issues of access to infrastructure and ethical guidelines. At the heart of it needs to be an AI-augmented teaching assistance and not a replacement.” She feels that the main legal challenges of AI-based learning systems are data protection and privacy, algorithmic biases and accountability. “Besides linguistic diversity, the Indian education system is diverse in pedagogy, infrastructure and geographical spread, demanding a focused approach that is not limited to local language tools,” she argues.

According to her, among the many challenges that AI learning poses is the risk of automation that depletes teacher roles, and teacher-student interaction that alters the learning environment. “The moves towards AI can also lead to homogenisation. The continuing inequalities from class, gender and caste in Indian classrooms need a humanised, empathetic pedagogy which technology-mediated environments cannot replace.” Her worries are similar to those expressed by, among others, Christian Terwiesch, a professor at the Wharton School (https://shorturl.at/jZfmG).

With OpenAI “actively hiring for roles in India” as part of a growth and survival strategy and the government being thrilled about it, it is only a matter of time before the company’s business interests bloom in the country.

And yet, Professor Banerjee’s warning about the “judicious” use of generative AI tools rings loud.

artificial intelligenceeducationlearningOpenAItechnology