Rahil Parikh, an Applied Scientist at Amazon, is spearheading groundbreaking research in privacy-preserving artificial intelligence (AI), a field crucial to safeguarding the personal information of millions of users and organizations worldwide. He is addressing one of the most critical challenges in modern AI: advancing AI capabilities while ensuring user privacy.
Parikh, a Birla Institute of Technology and Science (BITS) Pilani Goa Campus alumnus, has published several influential papers addressing critical privacy concerns in the age of large language models and AI-powered voice assistants. His research has been presented at prestigious conferences such as the Association for Computational Linguistics (ACL) and the Association for Computing Machinery (ACM) and has exposed vulnerabilities in natural language understanding models and proposed innovative solutions to mitigate these risks.
“As AI becomes increasingly integrated into our daily lives, ensuring user privacy is paramount,” Parikh explains. “My research aims to develop AI systems that maintain high performance while rigorously protecting sensitive data.”
One of Parikh’s key contributions is a novel technique to extract sensitive data on which publicly available large language models are trained on. He developed techniques to defend AI systems from such privacy attacks, preventing potential leakage of personal information embedded in AI, thus contributing to more secure AI systems. It has significant implications for voice assistants, healthcare applications, and financial services that increasingly rely on AI technologies. Parikh presented this work in Dublin, Ireland in 2022 and demonstrated his commitment to identifying and addressing potential privacy breaches in AI systems.
Parikh now applies his expertise to enhance the privacy features of products used by millions globally. His algorithms help safeguard user data in voice assistants and other AI-powered devices, setting new standards for privacy in consumer electronics.
“Our goal is to ensure that as AI becomes more powerful and pervasive, it remains a trustworthy technology that respects user privacy,” Parikh states. “This effort includes not only developing new algorithms but also shaping policies and standards for responsible AI development.”
Parikh’s research not only protects current users but also paves the way for more ethical and unbiased AI systems in the future. “AI is trained on data which tends to contain societal and historical biases. I want to build models that are fair and work equally well for all users, regardless of their gender, race, location or social status”, says Parikh. This is particularly crucial as AI applications expand into more sensitive areas such as enterprise tooling, healthcare, finance, and personal assistants.
Parikh’s journey from Goa to the forefront of AI research serves as an inspiration for aspiring technologists, showcasing the global impact of Indian talent in addressing critical challenges of the digital age. Before working in AI privacy, Parikh worked on developing AI systems to assist paralyzed patients in Italy and on building smart voice and speech-based assistants. His work demonstrates that it’s possible to push the boundaries of AI capabilities while simultaneously enhancing user’s interests and protections.
As AI continues to advance, the significance of Parikh’s work in privacy-preserving AI is paramount. It serves as a crucial safeguard, ensuring AI’s benefits can be realized without compromising privacy – a balance essential for responsible AI development in the future.
