Audrey Woods, MIT CSAIL Alliances | June 9, 2025
MIT CSAIL researcher Dr. Amar Gupta has spent decades pushing technological boundaries around the world, from his early career facilitating the computer revolution in India to inventing the technology behind check processing that has shaped modern banking. With a focus on creating tools that help the greatest number of people, Dr. Gupta has worked across international borders and academic boundaries to pull together ideas from economics, healthcare, management, and computer science and broaden the reach of computer science in realistic and applicable ways.
Now, with the collaborative support of Brazilian bank and CSAIL Alliances Affiliate Itaú, Dr. Gupta is exploring how large language models (LLMs) can be used to anticipate and detect anomalies such as fraud.
GETTING HIS START: INTERNATIONAL EXPERIENCE & FUTURE-FOCUSED THINKING
When Dr. Gupta was an undergraduate at the Indian Institute of Technology, he became fascinated by computers, which were still at that point relatively new. He wrote his senior thesis designing, implementing, and demonstrating a digital taxi meter to calculate fares, an entirely new concept at the time. After graduation, he joined the first company in India to produce pocket calculators and went on to work for the Indian government on setting up the first computer center for federal government agencies in the capital city; it is now called the National Informatics Center (NIC) and covers government offices across the country. Recognizing that computers would only become widespread in the country if people could maintain them, Dr. Gupta was instrumental in advocating for computer maintenance infrastructure and played a pivotal role in setting up the national corporation that could maintain the imported computers. After serving as a diplomat in London and starting his PhD research in India, he came to the United States to work at MIT.
With a doctoral degree in computer science and an MBA from the MIT Sloan School of Management (where he was the first person of Indian origin to hold a senior academic position), Dr. Gupta pursued problems that bridged computer science and business for real-world impact. One of Dr. Gupta’s most impactful innovations is the invention of the technology behind digital check reading and processing that most of us use on a regular basis. Dr. Gupta describes how, before such technology was available, over 50 billion checks were processed manually each year in the US alone, which, at an average processing cost of $1.25 per check, added up to billions in worldwide spending. Beyond such enormous cost and effort, there was also a widespread lack of standardization. In Britain there were no restrictions on how a check was written as long as it was in written form. Dr. Gupta shared a historical anecdote of a person in the UK who brought in a sheep with a check written on its side where the bank was forced to honor that transaction. At the time, Dr. Gupta was studying methods for digitizing textual documents for the Air Force and realized that technology could be developed to eliminate giant, wasteful, and slow systems around physical documents, including checks. Working with banks around the world, he developed an AI-driven way to digitally read written numbers with high accuracy and advocated for widespread standardization. His ideas weren’t implemented in the US until the catastrophe of 9/11 when planes were grounded and $60 billion worth of checks got stuck in the system over three days. “That's how my work got really serious consideration and got adopted,” Dr. Gupta says. Based on data from the Bank for International Settlements and the Federal Reserve, researchers have estimated that this technology has led to $1.28T of savings since its implementation.
This kind of prescience is a hallmark of Dr. Gupta’s career. In 2011, he wrote a paper criticizing the “overlapping, inconsistent, and inadequate legal and regulatory frameworks” around telemedicine, pointing out that “such barriers not only fail to advance public policy goals, but are unconstitutional when they restrict the practice of telemedicine across state and national borders.” While this paper received broad pushback when published, Dr. Gupta says it led to crucial conversations around policies and procedures that ultimately changed the legal landscape and enabled widespread telemedicine practices during the COVID-19 pandemic. He also co-developed early PC-based presentation graphics which helped shape modern clipart (an advancement that was recently recognized by the Computer History Museum in Mountain View, CA), proposed rotating global shift work across time zones to avoid graveyard shifts and reduce health risks, and was instrumental in securing funding and launching the Internet Telephony Consortium at MIT which helped legitimize internet calling worldwide, paving the way for services like WhatsApp and Zoom.
Recently, Dr. Gupta has turned his attention to another problem with global implications and potentially disastrous consequences: AI-enabled fraud.
CURRENT RESEARCH: LLM-ENABLED ANOMALY DETECTION
The rise of AI has many exciting elements, but financial institutions such as Itaú are all too familiar with the negative implications. “The fraud industry is growing at a very fast pace,” Dr. Gupta says. According to Nasdaq and the American Banking Association, fraud and bank scams accounted for $485.6 billion in global losses in 2023. This is exacerbated by AI tools such as LLMs, which empower even novice attackers with high-tech capabilities. Sponsored by Itaú Unibanco, the largest banking institution in Latin America, Dr. Gupta and his group are studying how AI can solve the problems it is being used to create via automatic defenses and rapid anomaly detection. “Trust is fundamental to our lives,” said Jair Antunes from Emerging Tech at Itaú Unibanco and leader of the MIT relationship. “Our collaboration with Dr. Gupta will bolster it not only for Itaú and the financial-services industry, but for society as a whole.” At the 2025 CSAIL Alliances Annual Meeting, Head of Emerging Technologies at Itaú Roberto Frossard said, “every single week we find something new with generative AI. Dr. Gupta has students working hard to deliver their best to us.”
One critical element of this research is Dr. Gupta’s insistence that everything be validated with classical methods as well as AI. Until there are better ways to control and predict hallucinations, Dr. Gupta believes strongly in double-checking everything and never fully relying on an LLM for accuracy. He and his team are also taking precautions against publicizing work that might aid bad actors, ensuring—to the best of their ability—that their efforts are as robust, reliable, and beneficial as possible.
Dr. Gupta finds Brazil to be one of the best countries to do research with, since “the people are serious, committed, and respectful.” Back when he was working on check processing and struggling to get banks to give him sample checks due to privacy concerns, Dr. Gupta describes how, when he asked a Brazilian bank for such sample data, “they said, ‘how many checks do you need?’ I told them two thousand, and they answered, ‘you’ll have them tomorrow morning.’” The managers had set up dummy accounts and had each employee fill out forty or so checks which Dr. Gupta could use to train his system. This kind of innovative spirit makes Brazil a natural and productive collaborator in Dr. Gupta’s research.
Beyond anomaly detection, Dr. Gupta is also working to address areas where he sees inequities or disparities, from medicine to mortgage loans. Previously, he and his team have exposed problematic biases, where a white male is more likely to be given a painkiller or a higher loan amount than a minority. “What we found is that the bulk of the data which is used for training relates to white males. This disparity in the training sample is what leads to the bias at the end.” So Dr. Gupta is exploring how to “develop AI technology which will better cater to people from diverse communities, diverse languages, and diverse specialties.” He is also continuing his work on digitizing data such as purchase orders, tax forms, driver’s licenses, etc. as well as exploring differential privacy to overcome challenges in data sharing between companies.
MENTORSHIP & MESSAGE TO INDUSTRY
Mentoring students has always been a core part of Dr. Gupta’s mission, particularly undergraduate students. He believes that the most innovative work comes from undergrads, a concept he describes as “fairly unique.” Freed by a relative lack of experience and knowledge of what’s possible or impossible, undergrads are “able to think about original ways to address problems.” He’s even published papers with high school students, which has led to not only success in those students’ lives but creative breakthroughs in the field. Dr. Gupta also favors the unusual approach of deploying several student teams to work on the same problem simultaneously, bringing forward multiple new ideas and solutions at once. Some of the students Dr. Gupta has mentored have had significant impact of their own, such as Salman Khan, founder of Khan Academy.
After a career watching computer technology become internationally ubiquitous, Dr. Gupta is concerned that “the way we are making some decisions—or not making some decisions—will come at a great cost to us.” This includes a lack of widely accepted procedures, which create disparities in regulation and unequal enforcement, and a reluctance to share mutually beneficial data such as emerging fraud methods. He worries that some of the costs of AI might only be explored and corrected by a “problem of some significant scale,” like 9/11 or COVID.
But Dr. Gupta has also seen firsthand how international cooperation and a spirit of transparency can lead to effective computer science solutions that change lives and improve systems for all. As governments and industries grapple with the challenges of AI, Dr. Gupta will continue to work on thoughtful, inclusive, realistic approaches that offer real-world value and solve everyday problems. For Dr. Gupta, the challenge isn’t just building better systems—it’s ensuring those systems work better for everyone.
Learn more about Dr. Gupta on his LinkedIn or CSAIL page.