Loading…
Thursday February 20, 2025 9:30am - 10:20am PST
Srik Gorthy, ByteDance (TikTok), Senior Data Scientist

In the era of Large Language Models (LLMs) and AI-driven insights, the quality of data holds the key to business decisions. Yet, biased samples — often overlooked or underestimated — can lead to flawed conclusions, unreliable models, and a loss of trust in data-driven systems. Whether biases stem from data collection methods, under-represented groups, or feedback loops in AI systems, understanding and addressing them is crucial for maintaining the integrity of insights.

This presentation examines the pervasive challenge of biased data samples and discusses key issues that data practitioners frequently encounter. From selection bias, where certain groups are systematically excluded, to response bias, where participation is skewed, biased sampling can distort the data landscape in subtle but significant ways. These biases have direct real-world consequences: for example, a cancer detection system that erroneously associates the presence of a ruler in mole images with malignancy, or an autonomous driving system trained without accounting for real-world deviations from traffic rules.

Recognizing bias is just the first step. The presentation will explore actionable strategies — like thoughtful sampling design and bootstrap resampling — to detect and mitigate bias, ensuring insights reflect reality more accurately. By examining real-world examples and practical approaches, this discussion invites professionals to rethink how data is collected and analyzed. Looking beyond the data means understanding its imperfections, challenging assumptions, and adopting techniques that uphold reliable decision-making in an increasingly AI-centric world.
Speakers
avatar for Srik Gorthy

Srik Gorthy

Experienced Data Forager, ByteDance (TikTok)
Srik Gorthy has over a decade of experience in Machine Learning and Data Science, having worked in and contributed to multiple projects across industries like internet technologies, semiconductors, and FMCG. Currently, as a Senior Data Scientist at ByteDance, he leads data science... Read More →
Thursday February 20, 2025 9:30am - 10:20am PST
VIRTUAL Frontend World https://app.events.ringcentral.com/events/developerweek-productworld-ai-devworld-2025/reception

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Share Modal

Share this link via

Or copy link