Expressions, Tone, Gestures: Building Blocks of a Personal AI Agent

Genesislab CEO Young-bok Lee told IT Chosun (in its “AI Leaders 2024” series) that “AI interview tools are the starting point of the future AI agent.” The facial-expression, tone, and gesture analysis used in hiring, he argued, is the path to accumulating the kind of data that a super-intelligence era will run on.

Lee describes the behavioral data ViewinterHR captures in a one-minute interview video as “an individual’s unique dataset” — uncommon in form, and decisive material for building a personalized AI agent. ViewinterHR is already being used not only for hiring but also for training and promotion assessments, which means moments of evaluation around a single person are being digitized across multiple contexts. “When these individual agents are aggregated, they become one axis for building super-intelligence,” Lee said.

The other axis is trust. Genesislab breaks it into three components: fairness (bias minimization), validity (accurate evaluation), and transparency (interpretability). In customer deployments of ViewinterHR, the correlation coefficient with human evaluators has come in above 0.7 — roughly twice the 0.35 threshold the U.S. Department of Labor uses. As AI legislation takes shape across Korea, the U.S., and Europe, Lee expects this trust indicator to become a market-entry condition in itself.

Source: IT Chosun (IT조선) — Genesislab CEO Young-bok Lee: “AI Interview Tools Are the Starting Point of the Future AI” (AI Leaders 2024)