AI Can Generate Answers, But Can We Trust Them? | Arun Prem Sanker | TEDxCNS Pune Youth
Description: Artificial intelligence can generate code, analyze data, and produce answers in seconds. But how do we know whether those answers are correct, meaningful, and useful in the real world?
In this talk, Arun Prem Sanker explores one of the biggest challenges in building reliable AI systems: evaluating whether their outputs can be trusted. He explains why a technically correct answer may still lead to the wrong decision when an AI system lacks the business context behind the data.
Drawing from his experience in product data science, Arun examines how automated evaluation, human oversight, and continuous feedback can work together to make AI systems more dependable. He also challenges us to look beyond whether an AI system produces an answer and ask whether that answer actually helps someone make a better decision.
As AI becomes increasingly involved in consequential decisions, his message is clear: trust cannot simply be assumed. It must be tested, measured, and continuously earned.
Because generating knowledge is only part of the challenge. Making sure it holds up in the real world is what gives it value. Arun Prem Sanker is a product data scientist and technology leader with extensive experience in data science, machine learning, experimentation, and product analytics. His work focuses on building data-driven decision systems, designing reliable experimentation frameworks, and using advanced analytics to understand user behaviour and improve product performance. He is passionate about bridging the gap between technology and real-world outcomes, helping organisations make informed decisions through data, evaluation, and innovation. This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at https://www.ted.com/tedx Close description
In this talk, Arun Prem Sanker explores one of the biggest challenges in building reliable AI systems: evaluating whether their outputs can be trusted. He explains why a technically correct answer may still lead to the wrong decision when an AI system lacks the business context behind the data.
Drawing from his experience in product data science, Arun examines how automated evaluation, human oversight, and continuous feedback can work together to make AI systems more dependable. He also challenges us to look beyond whether an AI system produces an answer and ask whether that answer actually helps someone make a better decision.
As AI becomes increasingly involved in consequential decisions, his message is clear: trust cannot simply be assumed. It must be tested, measured, and continuously earned.
Because generating knowledge is only part of the challenge. Making sure it holds up in the real world is what gives it value. Arun Prem Sanker is a product data scientist and technology leader with extensive experience in data science, machine learning, experimentation, and product analytics. His work focuses on building data-driven decision systems, designing reliable experimentation frameworks, and using advanced analytics to understand user behaviour and improve product performance. He is passionate about bridging the gap between technology and real-world outcomes, helping organisations make informed decisions through data, evaluation, and innovation. This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at https://www.ted.com/tedx Close description