iiinova — where your digital self keeps evolving

Maty Bohacek

@matybohacek.bsky.social

68 followers 249 following 4 posts

AI @ stanford www.matybohacek.com

Posts Replies Media

Reposted by Maty Bohacek

Christo Buschek @handle.invalid · 18/05/2026

Benchmarks as narrative devices by AI builders! Our new paper maps 231 benchmarks highlighted in 139 model releases by 11 model builders in 2025. I'm excited to present “Unsteady Metrics and Benchmarking Cultures of AI Model Builders” next month at #FAccT2026. Paper at arxiv.org/abs/2605.14164

The abstract of the "Unsteady Metrics and Benchmarking Cultures of AI Model Builders" paper, highlighting some findings.

172

Reposted by Maty Bohacek

APAI @ ICCV 2025 @apaiiccv25.bsky.social · 13/06/2025

📣 Introducing APAI 2025 — Authenticity & Provenance in the Age of Generative AI Workshop 📅 This October in Honolulu, HI 🇺🇸 📍 In-person @iccv.bsky.social Stay tuned for speaker and challenge announcements! 🔗 sites.google.com/berkeley.edu...

sites.google.com

APAI @ ICCV2025

OVERVIEW Generative AI allows for the rapid and automatic generation of highly realistic audio, images, and videos (so-called deepfakes). The field of media forensics and digital provenance focus on d...

132

Maty Bohacek @matybohacek.bsky.social · 18/05/2025

Great work by WITNESS — we need more benchmarks informed by real-world impact like this: arxiv.org/abs/2504.21489

arxiv.org

TRIED: Truly Innovative and Effective AI Detection Benchmark, developed by WITNESS

The proliferation of generative AI and deceptive synthetic media threatens the global information ecosystem, especially across the Global Majority. This report from WITNESS highlights the limitations ...

010