Indian policymakers specialised in artificial intelligence (A.I.), who were surveyed by tech giant OpenAI on AI risk perceptions broadly said that threats to education from A.I. are a uniquely high area of concern in India compared to other countries. However, experts within government, industry, and academia told The Hindu that they disagreed with many findings of the OpenAI risk perceptions study. They said that A.I. threats to education are overblown, misplaced, and failed to recognise that the benefits greatly outweigh the dangers in India.
OpenAI’s private research conducted between September to December of 2023 through surveys and expert interviews with a few dozen policymakers in five countries found that “Education risks (e.g., students over-relying on AI tools at the expense of critical thinking skills), were viewed as least risky,” but “India is a notable exception: Indian respondents ranked risks to education as the fifth priority area of concern, greater than geopolitical risks or the alignment problem.”
No explanation was given in the OpenAI study for why Indian policymakers found A.I. in education risks to be of particularly high concern. OpenAI did not respond to multiple requests for comment for this article.
The OpenAI study, which The Hindu exclusively obtained, focused on four broad categories: benefits and risks from A.I., pace of A.I. development, AGI (Artificial General Intelligence) and existential risks, and A.I. risk management. The study implicitly focused on cutting edge generative A.I. use cases such as A.I. tools that generate new text, images, videos etc rather than broad uses of artificial intelligence which have existed for many years.
OpenAI, the largest and most popular generative A.I. company in the world, found in the study that the greatest dangers from the technology came from “ ‘A.I. misuse/malicious use’ by bad actors and ‘economic risks’ (like job displacement due to automation),” according to policymakers surveyed in five different countries: India, Japan, Taiwan, U.K., and the U.S.
On the other hand, OpenAI’s risk perceptions study, which was not released publicly, found that “advanced research and discovery and health advancements,” were identified by survey respondents from all countries as the most beneficial applications of A.I. over the next five years.
OpenAI’s private study surveyed 47 policy officials including those within the Indian government, civil society, and A.I. scholars, with a focus on career officials within the Ministries along with the non-governmental entities in India.
Advitiya Sharma, chief growth officer of leading EdTech company Schoolnet, disagreed with parts of the OpenAI study findings, highlighting the fundamental issue with generative A.I. use in particular within education in India is the lack of understanding and awareness which creates a lack of confidence and overstates the problems associated with the technology.
“We’ve already worked with tens of thousands of teachers and students who have used A.I. in their schools and the benefits outweigh the risks or dangers that exist,” said Sharma who is a serial entrepreneur who previously co-founded Housing.com and founded Genius Teachers, a quiz-based e-learning platform.
“The other four countries in the OpenAI study are much more ahead of India regarding their use and understanding of A.I. in education and other fields so there’s bound to be a higher perception of problems in India if it’s not been used much before,” Sharma said.
Sharma added that if the OpenAI study were conducted again in 18-24 months, the risk perception of A.I. in education in India would drastically decrease due to widespread use and awareness of the benefits of it as well as the vulnerabilities going down.
Some of the specific elements of A.I. policymaking in India that OpenAI conducted its survey about include pitfalls within the education sector as well as the growing optimism and trust that India has placed in public-private cooperation mechanisms.
The potential loss of critical reasoning among students due to the use of the technology in education in India is one of the most important considerations and threats, according to Indian educators and policymakers surveyed by OpenAI.
For example, thanks to cutting-edge generative A.I. tools, a student may not need to think or reason out a problem because the technology can give them an immediate answer and thereby could cause a decrease in original thinking and problem-solving skills among students.
Another potential worry of Indian educators is that if students around the country are using generative A.I. tools like OpenAI’s popular ChatGPT chatbot then schools could face the challenge of testing machine intelligence rather than the students themselves and this could force educators to change the way they assess and test students.
“Some challenges definitely exist like when A.I. tools produce incorrect or misleading results but still I’m very optimistic about A.I. in education because it’s great for brainstorming and is a very useful knowledge co-creation tool,” said Krishnan Narayanan, Research Lead at IIT Madras’ Centre for Responsible AI, which is driving a ‘GenAI4Edu’ initiative.
“We just have to learn how to design A.I. systems to be Socratic, to get the technology to force students to come up with or be guided to answers themselves rather than giving them the answer automatically,” said Krishnan, an award-winning author of ‘Against All Odds: The IT Story of India’, and former senior executive at Infosys.
There are particularly low levels of engagement in learning and education among the vast majority of students in India, according to multiple experts who spoke to The Hindu. Most children fail to do their homework or engage substantively with their course material.
This atmosphere creates a unique opportunity for cutting-edge technologies like generative A.I. tools to improve learning outcomes and help students interact positively with their study material, said former Rajya Sabha MP and former BJD party IT and Tech cell head, Amar Patnaik.
“A.I. in education is a boon in disguise because it will force and push students to have critical thinking skills and creativity which is sorely lacking in India,” said Patnaik.
“The OpenAI study says that A.I. in education could negatively affect students but I think instead that it will make more students interested in learning and therefore school curriculum will need to be changed significantly because of A.I. tools that everyone will and should use,”Patnaik added.
Furthermore, some educators say that the technology will force students to learn in a digital-friendly and evidence-based way in the near future.
“A.I. systems can be beneficial in education because the more you put into the prompt, the more you explain your understanding of a problem, the better your answer will be from A.I. tools. So we have to see this as an opportunity for educators to use the technology to get students to engage more on subjects that they’re not always interested in,” said Krishnan from IIT Madras.
Another potential benefit of A.I. educational tools in India is that they could provide a better, more productive outlet for students to spend their free time, especially in comparison to social media platforms.
“With kids spending so much time on their devices and computers, A.I. systems provide a far, far better alternative than social media or other things they do on their devices. It’s far more fruitful for students to use fun educational A.I. tools like ours compared to spending an hour on TikTok or Instagram,” said Sharma from Schoolnet.
Nevertheless, educators, politicians, and tech executives say that some thoughtful, carefully considered guardrails and regulations will be needed to successfully implement and use the technology in the long run.
“Guardrails for A.I. in education to ensure it’s accurate, unbiased, and helpful have to be taken seriously because students are our future and we need to ensure that they learn the right way,” said Krishnan.
“If you don’t expose students to thoughtfully designed generative A.I. educational tools that use a Socratic style and have other checks and balances they will be forced to learn from general A.I. tools and the internet at large which have many existing issues as we know.” Krishnan added.
Another challenge is ensuring that the data collected in India to create and run A.l. tools within the educational sector as well as other fields are appropriately gathered and that students don’t blindly follow the output of the technology.
“A.I. biases baked into the current data sets are a big issue. So if you’re making decisions based on flawed or biased data sets and their corresponding A.I. tools, there will be worse or problematic actions taken,”said Patnaik, the former Rajya Sabha member.
“In the Indian context, as a civil servant and public official, I have the experience to know that we cannot create A.I. systems that are built only upon data collected through formal official channels such as government records, consumer purchases, online behavior, etc. because a lot of insights and answers to problems in India come through informal data channels from entities like street vendors, labourers, unskilled workers, and others in informal sectors,” Patnaik added.
Although potential A.I. risks to education in India were highlighted in the OpenAI study, major concerns around other prominent use cases of the technology deserve just as much or more attention in the coming months, according to some tech-policy experts like Krishnan from IIT-Madras and Patnaik, the former Rajya Sabha member.
Krishnan said he was surprised that the results of OpenAI’s risk perception survey didn’t highlight threats like the role of A.I. in the military and warfare while Patnaik said he would have expected the study to highlight how the technology could be used negatively or positively to impact climate change.
EdTech executives like Sharma from Schoolnet say that the future of A.I. in education is full of excitement and innovation because of the potential to democratise and scale access to world-class educational tools across India including among the poorest schools and students.
“Soon we will have an A.I. avatar or teacher trained on Einstein, Aristotle, Tagore, you know the best of the best, and every student then gets access to a genius teacher,” said Sharma.
“For example, if a student gets stuck on a problem or has a complicated question, they can turn to an A.I. chatbot that gives outputs similar to that of Einstein or other world-renowned thinkers. We have such an A.I. application currently in an early beta stage of testing but has been rolled out with some success in a few Jharkhand government schools and of course the more users and the more data you feed the technology, the better it gets.”
Published – December 26, 2024 03:39 pm IST
