How Does Your Phone Know Your Face?

You glance at your phone and it unlocks — no password, no fingerprint. But your phone never sees your face the way you do. It sees a list of numbers. From 1987 ghost faces made of pure math, to Apple's 30,000 invisible infrared dots, to the uncomfortable 2018 study that exposed who this technology fails — this is the story of how machines learned to recognize a human face. CHAPTERS 00:00 The Hook 00:50 The Ghost Faces of 1987 02:22 Detect, Align, Embed, Match 04:05 The Phone That Shoots Lasers at Your Face 05:38 The Uncomfortable Numbers 08:08 The Battle Against Fakes 09:16 Your Face, As Data SOURCES Scholarpedia, "Eigenfaces" (Sirovich & Kirby 1987; Turk & Pentland 1991, MIT Media Lab): http://www.scholarpedia.org/w/index.php?ti... TheFreeDictionary encyclopedia, "Eigenface" (history): https://encyclopedia.thefreedictionary.com... PPC Land, "Explaining facial recognition" (DeepFace 2014 four-stage pipeline; FaceNet 2015 128-dim embedding): https://ppc.land/facial-recognition/ BytePlus, "What is Facial Recognition?" (detection, alignment, embedding, matching, thresholds): https://www.byteplus.com/en/what-is/facial... 9to5Mac, Apple unveils iPhone X (TrueDepth, A11 Bionic neural engine, 2017): https://9to5mac.com/?p=499074 NFCW, Apple demos Face ID (30,000+ IR dots, secure enclave, on-device): https://www.nfcw.com/2017/09/12/354620/app... Bustle, "What Is Face ID?" (1 in 1 million vs 1 in 50,000 Touch ID): https://www.bustle.com/p/what-is-face-id-h... iFixit, iPhone X teardown (flood illuminator, dot projector, IR camera): https://www.ifixit.com/Guide/diff/step/989... Communications of the ACM, "Can the Biases in Facial Recognition Be Fixed; Also, Should They?" (Gender Shades 2018: 34.7% vs 0.8%; NIST FRVT 2019: 10-100x false positives): https://cacm.acm.org/magazines/2021/3/2506... The Science Story, "Is facial recognition too biased to be let loose?" (NIST FRVT 2019, convolutional neural networks accuracy leap): https://www.thesciencestory.com/wp-content... #facialrecognition #faceid #artificialintelligence #machinelearning #technology #computerscience #neuralnetworks #biometrics #documentary #stem #privacy #datascience Voiceover: This video uses an AI-generated voiceover.