Evidence Over Impression: Building AI Hiring Tools Your Team Can Actually Trust
A practitioner's framework for evaluating AI interview tools — grounded in compliance, evidence-based design, and real-world lessons from building in the space.
Your hiring team interviews 15 candidates per open role. Maybe 4 are actually the right fit. That's 70% of your team's interview time producing weak or misleading signal. The problem isn't bad candidates — it's that the first meaningful evidence about a candidate often arrives too late, after your team has already spent their bandwidth. AI tools promise to fix this. But most AI hiring solutions today either give you static video questions with no real intelligence, or opaque scoring systems that create more compliance risk than clarity. In a regulatory environment where NYC, California, Illinois, and Colorado all now have active or imminent AI hiring laws — and where the EU AI Act classifies recruitment AI as high-risk — the stakes for getting this wrong have never been higher. In this session, Sumanta Chakraborty — a 25-year technology veteran who led AI product teams at NVIDIA and Nuance before founding an AI hiring intelligence company — shares the practical frameworks her team developed for building AI interview systems that are evidence-based, explainable, and compliant by design. You'll learn why "evidence over impression" isn't just a product philosophy but a legal and ethical imperative, how to evaluate whether any AI hiring tool meets the emerging compliance bar, and what "human-in-the-loop" actually needs to look like in practice. This is not a vendor pitch. It's a practitioner session from someone building in the space who believes HR professionals deserve a clear-eyed framework for navigating AI in hiring — one that protects candidates, satisfies regulators, and actually helps your hiring managers make better decisions.
Learning Objectives (3–4): Apply a compliance checklist across NYC LL144, California's Civil Rights Council regulations, Illinois HB 3773, and Colorado's AI Act to evaluate whether current or prospective AI hiring tools meet emerging legal standards. Use the "Evidence Over Impression" framework to distinguish defensible, transcript-backed hiring evidence from vague impressions, personality scoring, and problematic behavioral proxies — and understand why this distinction matters for both decision quality and legal defensibility. Design meaningful human oversight structures where hiring managers can interrogate AI-generated evidence, identify gaps, and make informed decisions — rather than rubber-stamping opaque AI recommendations. Identify red flags in AI hiring vendor claims by asking the right questions about signal design, explainability, and bias audit readiness — protecting your organization from tools that create more risk than value.
About Sumanta Chakraborty ('10) — Chief Executive Officer/ Co-Founder of nSpire AI
Sumanta Chakraborty is the founder and CEO of nSpire AI. Before starting nSpire, she spent 25+ years building products and technologies across companies like NVIDIA, Nuance, Sony, LG, and Siemens — working on systems spanning smartphones, robotics, autonomous vehicles, and speech/NLP platforms. Over time, she noticed a consistent gap: talent is everywhere, but access to opportunity and meaningful guidance is not. nSpire AI was built to address that. They're developing the intelligence layer for the future of hiring — with Theo, an AI career coach helping candidates improve through structured practice and feedback, and Synthia, an AI-powered system that helps companies make better hiring decisions through evidence-based evaluation. Their mission is simple: make personalized career support and better hiring decisions accessible at scale.