AI Tools for Trust: Community Notes, Rhetoric Detection & More
forethought.org · 4,651 words · saved by 2 readers
Five AI technologies to combat misinformation: community notes, rhetoric detection, reliability tracking, epistemic evals, and provenance tracing.
Design Sketches: Collective Epistemics 25th December 2025 [AI Narration] Design Sketches: Collective Epistemics Playback speed Volume 0:00 of 33:35 Introduction We think that near-term AI could power tech that could be a big deal for how we collectively share and process information, and work out what to trust. We want to help people envision this. In this piece, we will sketch five potential technologies that might help, explaining how they could work, and what might be achieved in a world that adopts them: Community notes for everything — Anywhere on the internet, content that may…
saved by
related reading
- [2503.03750] The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systemsarxiv.org
- [2510.17941] Believe It or Not: How Deeply do LLMs Believe Implanted Facts?arxiv.org
- Believe It or Not: How Deeply do LLMs Believe Implanted Facts?alignment.anthropic.com
- AI Tools for Existential Securityforethought.org
- AI Resilienceairesilience.net
- Deep Deceptiveness — LessWronglesswrong.com
- The Compute Verification Postfirstscattering.com
- [2004.07213] Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claimsarxiv.org
- the case for CoT unfaithfulness is overstated — LessWronglesswrong.com
- Ten AI safety projects I'd like people to work onthirdthing.ai
- What Humanity Needs To Flourish In The Next Decadenoemamag.com
- How AI Is Learning to Think in Secretnickandresen.substack.com