Shanthi
Sivakumar
Founder and AI Ethics Consultant at FairFrame AI Org
"Designing for Intelligence Why High Performing AI Systems Still Fail Real People"
"Designing for Intelligence Why High Performing AI Systems Still Fail Real People"
Tue May 12 - 5:40 PM EDT/New York (See in local time)
Add to Calendar
05/12/2026 5:40 PM
05/12/2026 06:00 PM
America/New_York
#WTGC2026
"Designing for Intelligence Why High Performing AI Systems Still Fail Real People"
#WTGC2026
"Designing for Intelligence Why High Performing AI Systems Still Fail Real People"
https://www.womentech.net/ringcentral
https://www.womentech.net/ringcentral
"Designing for Intelligence Why High Performing AI Systems Still Fail Real People"
"Designing for Intelligence Why High Performing AI Systems Still Fail Real People"
Don’t miss out and join visionaries, innovators, and thought leaders from all over the world at the Women in Tech Global Conference.
Vote by Sharing
Unite 100 000 Women in Tech to Drive Change with Purpose and Impact.
Do you want to see this session? Help increase the sharing count and the session visibility. Sessions with +10 votes will be available to career ticket holders.
Please note that it might take some time until your share & vote is reflected.
Session: Designing for Intelligence Why High Performing AI Systems Still Fail Real People
Most conversations about AI bias start in the wrong place. We argue about model accuracy and training data, while the real damage is already done much earlier in the data platform itself.After a decade designing and delivering enterprise and healthcare data systems, I have seen this pattern repeatedly. Systems that look intelligent at scale quietly fail the same users again and again. Not because the model is wrong, but because of everyday design decisions no one revisits. Which data arrives first. What gets aggregated away. Which workflows are optimized for speed instead of clarity. Whose friction is treated as noise.
Bias in AI behaves much like advanced driver assistance systems in modern cars. They work almost all the time. They pass validation. And yet they consistently miss edge cases that require judgment, context, or courtesy. These failures rarely show up as errors. They surface as outliers, escalations, and user frustration that teams learn to ignore.
This talk focuses on the human AI interaction layer of data platforms where bias actually enters. Using real examples from high-stakes healthcare and enterprise systems, I will show how ingestion prioritization, default metrics, and dashboard design quietly shape decision making long before any model is trained.
Attendees will leave with practical, low-overhead techniques to evaluate AI UX fairness and user intent without slowing delivery. I will also share how these repeated blind spots led me to create FairFrame AI not as an ethics initiative, but as a practitioner response to failures I kept seeing in production systems.
This session is for teams building intelligent systems who want AI that earns trust in real workflows, not just in accuracy reports.
Key Takeaways
- Where bias enters AI systems before any model is trained
- How human-AI interaction design quietly shapes trust and decisions
- Why outlier user experiences are signals, not noise
- Practical, low-overhead ways to evaluate AI UX fairness
- How to design intelligent systems that work for real people, not just dashboards
Bio
Shanthi Sivakumar is a digital health innovator, AI ethics advocate, and founder of FairFrame AI, a nonprofit initiative focused on mitigating algorithmic bias in healthcare, education, and career technologies. With over a decade of experience in healthcare IT and consulting for Fortune 500 companies, Shanthi brings a rare blend of clinical insight, technical strategy, and ethical foresight to the development of responsible AI systems.