Q: Why Do Keynote Speakers Keep Suggesting That Improving Security Is Possible?
summary
Argues that machine learning is often opaque yet rolled into mission‑critical systems, cites criminal‑justice risk scores that label Black defendants twice the rate as high‑risk, facial‑gender detectors that hit white men at ~1% error but darker women at 20–35% error, and Microsoft’s Tay chatbot that learned racist output from live tweets; shows IoT medical devices sending cleartext instead of TLS; urges skepticism, better testing, and slower, audited deployment.
AI-written summary · about
chapters (44)
- 0:00Introduction
- 0:20Keynote Definition
- 3:39Value of Skepticism
- 4:06Machine Learning
- 13:37Egg Drop
- 14:27Other Winners Curse
- 16:22Two Mistakes
- 17:01Kingsley
- 19:44AI Mission Creep
- 20:33AI Workflow
- 21:05Conversation
- 22:56Tay
- 23:20AI Child
- 24:00Internet People
- 25:20Twitter Answers
- 25:45Why Is That
- 26:03Race War
- 26:33Grammar Lessons
- 26:59Prejudice Science
- 27:35Sotae Rant
- 28:17A Holistic View of Security
- 28:53Cartesian Security
- 29:19ARPANET
- 30:31Cartesian Approach
- 31:13Gadgets Are Eternal
- 32:09The Liberal Arts Are Getting Upset
- 32:54Theyve Made Mistakes Too
- 33:39Machine Learning In The Criminal Justice System
- 34:18Gender Detection And Faces
- 34:37White People Have Faces
- 35:22Flashlight Apps
- 36:49Technological Manifest Destiny
- 38:15Statistics
- 40:58Internet of Things
- 41:18IOT Security
- 42:01TLS
- 42:39Cheapness
- 43:39Best Seller
- 44:28Have I written a paper about IoT
- 45:52Why wouldnt you do this
- 46:46Avoidable Mistakes
- 47:18Conclusion
- 48:07Questions
- 49:23Would you use machine learning
related talks
44mAdvancing Cybersecurity with AI and Machine Learning: Opportunities and Challenges- 44mAdvancing Cybersecurity with AI and Machine Learning: Opportunities and Challenges
23mKeynote: Fireside Chat with Window Snyder & Katherine Druckman
01h 05mNavigating the Security and Privacy Landscape of Modern AI
01h 07mKeynote Presentation Blinded by “the weakest link” A New Perspective on Human Risk
28mSecurity of Machine Learning
49mCXO Panel: Cybersecurity: Is Machine Learning The Answer?
45mSecuring AI: There Is No Try, Only Do!