AI, truth, and the new information environment [Keynote]

on YouTube ↗48m

" summary

Claim: over 95+% AI-generated content online by 2025, producing a feedback loop where LLMs train on LLM output. Demonstrated specifics: Narendra Modi uses AI-translated speeches (approved), Taylor Swift nudes and Pelosi slowdown are unapproved manipulations, viral flood photos and the "shrimp Jesus" meme show scale and misattribution. Proposed responses: use AI for verification, demand probabilistic transparency about provenance, raise platform liability for algorithmic amplification, and teach information literacy to triage volume versus vetted quality.

AI-written summary · about

" chapters (12)

1. 0:00Introduction
2. 9:49AI and truth discussion
3. 10:09Truth in digital imagery
4. 19:25History of truth
5. 26:39Digital sharing flood
6. 28:23AI misinformation
7. 31:41Discerning AI truth
8. 37:43Spurious correlations
9. 41:12Incentives for misinformation
10. 44:24Adapting to new info age
11. 45:49Solutions and future
12. 48:18Closing and Q&A

" related talks

• 54mAI and Ethics: Navigating the Future of Responsible TechnologyScott Davis · Great International Developer Summit 2025
• 28mEthical AI?!Laura Summers, Vilas Dhar, Mehran Sahami, Navrina Singh · DLD Munich 23
• 37mThe Shadows That Follow AI Generative ModelsCheuk Ting Ho · Web Summer Camp 2023
• 22mAI and Disinformation: Sharpen Your Bullshit Detector!Marie Viley · BDX I/O 2025
• 22mAI & DemocracyLisa Witter, Robyn Scott, Azeem Azhaar · DLD Munich 24
• 36mDisinform your Surroundings: AI and disinformation campaignsTess (killrbunn3) · BSides Edmonton 2024
• 01h 08mAI and Elections Disinformation, Deepfakes, DystopiaCornelia Kutterer, Colin J Bennett, Maria Villegas Bravo, Elinor Wahal · CPDP 2024
• 35mDisinform your Surroundings: AI and disinformation campaignsTess (killrbunn3) · BSides Edmonton 2024