iiinova — where your digital self keeps evolving

Tzvi Ganel

@tganel.bsky.social

Tzvi Ganel @tganel.bsky.social · 10/07/2026

Humans and AI models are both benchmarked on facial age estimation with a single number: mean absolute error. Until now there's been no framework for what that number actually measures. Our new paper, open access in Scientific Reports, provides one.

120

Reposted by Tzvi Ganel

Mel Goodale @action-brain.bsky.social · 03/09/2025

How good are we at estimating someone's age from their face and how does that change with the face's chronological age? @tganel.bsky.social et al.'s new framework disentangles bias and error, revealing new insights into how facial age is computed by humans and AIs. 🧪 www.biorxiv.org/content/10.1...

Photos of the faces of three women, one young, one middle-aged, and one elderly. Text: Apparent facial age plays an important role in social interactions and is a meaningful marker of biological aging. Although both humans and AIs achieve reasonable accuracy in estimating age from a person’s face, performance remains imprecise. Drawing on principles from classical psychophysics, we show that the conventional measure used to benchmark the accuracy of human and AI performance is fundamentally confounded by response bias. We introduce a new framework based on simulated data, reanalysis of existing data, and new experimental results. Our framework provides novel insights into how facial age is processed by humans and AIs, offering new directions for future research and applications in the study of aging.

062