Links for 2023-05-04
Learning Physically Simulated Tennis Skills from Broadcast Videos
We present a system that learns diverse, physically simulated tennis skills from large-scale demonstrations of tennis play harvested from broadcast videos...our approach can learn complex tennis shotmaking skills and realistically chain together multiple shots into extended rallies...our system produces controllers for physically-simulated tennis players that can hit the incoming ball to target positions accurately using a diverse array of strokes...our system can synthesize two physically simulated characters playing extended tennis rallies with simulated racket and ball dynamics.
Project page: https://research.nvidia.com/labs/toronto-ai/vid2player3d/
Google DeepMind CEO Says Some Form of AGI Possible in a Few Years [Wall Street Journal] https://archive.is/s1tVY
The amazing AI super tutor for students and teachers https://www.ted.com/talks/sal_khan_the_amazing_ai_super_tutor_for_students_and_teachers/c
Most teachers, and many students, are already using ChatGPT for their job. https://www.waltonfamilyfoundation.org/learning/teachers-and-students-embrace-chatgpt-for-education
Mass Editing Memory in a Transformer https://memit.baulab.info/
Drone delivery by Zipline in Utah https://www.reddit.com/r/Utah/comments/13047hn/intermountain_healthcare_and_a_zipline_drone/
Multi-Agent Simulated Environment: Petting Zoo https://python.langchain.com/en/latest/use_cases/agent_simulations/petting_zoo.html
“We show that compiler intermediate representations (IRs) can be used to pre-train better multilingual models for code, resulting in a 11% increase in performance for unsupervised code translation across four languages…” https://twitter.com/b_roziere/status/1652900010013388801
ArK: Augmented Reality with Knowledge Interactive Emergent Ability https://augmented-reality-knowledge.github.io/
A model that uses human prompts and sketches to generate realistic fashion images https://techxplore.com/news/2023-04-human-prompts-generate-realistic-fashion.html
QuiLLMan: Voice Chat with LLMs https://github.com/modal-labs/quillman
Nanowire networks learn and remember like a human brain https://www.sydney.edu.au/news-opinion/news/2023/04/24/nanowire-networks-learn-and-remember-like-a-human-brain.html
AI-driven robots start hunting for novel materials without help from humans [Science] https://archive.is/oohr0
New laser technology developed by EPFL and IBM https://actu.epfl.ch/news/new-laser-technology-developed-by-epfl-and-ibm/
Researchers 3D print a miniature vacuum pump https://news.mit.edu/2023/3d-printed-miniature-vacuum-pump-0425
Institutional review boards save a few people while killing 10,000–100,000 more and costing $1.6B https://astralcodexten.substack.com/p/book-review-from-oversight-to-overkill
“Mental Models Of People Can Be People...In general I just don't create mental models of people anymore, and would recommend that others don't either.” https://www.lesswrong.com/posts/9dB54pBuDCkoedCSE/mental-models-of-people-can-be-people (See also this 2009 Eliezer Yudkowsky Q&A https://www.youtube.com/watch?v=lbzV5Oxkx1E)
The "run-and-tumble" motion of bacteria as a biological analogue for gradient descent? https://www.lesswrong.com/posts/iN7uwrBEFRxa5rjoL/natural-selection-vs-gradient-descent?commentId=KmW59hvh8T6krm7As
The beginning of the end of the password: “Unlike passwords, passkeys can only exist on your devices. They cannot be written down or accidentally given to a bad actor. When you use a passkey to sign in to your Google Account, it proves to Google that you have access to your device and are able to unlock it. Together, this means that passkeys protect you against phishing and any accidental mishandling that passwords are prone to…most people will find it easier to control access to their devices rather than maintaining good security posture with passwords…” https://security.googleblog.com/2023/05/so-long-passwords-thanks-for-all-phish.html
If you take this line of reasoning to its logical conclusion you get moral realism: https://slatestarcodex.com/2018/04/01/the-hour-i-first-believed/
Tweet: https://twitter.com/anderssandberg/status/1653478613151301651