Transparency & Control for Personalised AI
summary
TTC Labs @ Meta synthesizes 16 user studies plus expert co‑design sessions (15–25 experts per session in Brussels, Paris, Berlin, Dublin) and finds personalized AI delivers relevance, speed, connection by recalling history, calendars and preferences, but creates concrete harms: factual inaccuracy, privacy risks from broad data fusion, and agentic overreach when models act on inferred goals. Design levers that work are in‑context transparency, why am I seeing this retrospectives, progressive disclosure, and new controls for voice‑first and expanding memory scope.
AI-written summary · about
chapters (7)
- 0:00Transparency and control in AI
- 2:27Law, policy and technology co-design
- 5:54Personalization in AI experiences
- 11:11Privacy challenges in personalized AI
- 14:20User control and transparency design
- 18:10In-context transparency
- 20:56Voice-first AI future
related talks
35mDesigning Inclusive and Accessible Experiences in the Era of AI
18mDeveloping AI in a privacy conscious world
01h 17mPrivacy and New Technologies
46mTowards Trustworthy AI: The European Approach
54mAI and Ethics: Navigating the Future of Responsible Technology
31mA more human-centered approach to AI
45mAI, Corporate Responsibility & Democratic Legitimacy: Extended Q&A
21mEthics in the age of AI: Strategies for mitigation and their historical context