So funny that even three years ago “ML can’t do symbols” was a widespread belief. That belief has been completely trashed.
@reza_zadeh
-
ML Models Successfully Solving IMO and ICPC Competitions
By
–
With IMO and ICPC solved by ML models, this is rapidly becoming true
-
SFO Airport Excellence: Fast Security and Reliable Flight Access
By
–
SFO is the best US airport experience as a passenger. @flySFO Even showing up to the airport when boarding has already started, with Clear+Precheck+
@United app I can reliably make the flight. Now even better with Waymo. -
AI Fundamentally Based on Calculus, Linear Algebra, and Statistics
By
–
The funny thing is AI is mostly calculus underneath linear algebra and stats It's calculus all the way down
-
Waterloo Co-op Experience Led to Google and Language Models
By
–
Happy to chat. Waterloo coop was the reason I got into Google in 2005 working on Language Models.
-
Simple Adversarial Attacks in Real-World and Military Contexts
By
–
Simple adversarial attacks in the wild, in war
-
Ukrainian drone swarms destroy Russian bombers; DJI militarization concerns
By
–
The Ukrainians cleverly parked trucks with roofs filled with swarms of drones near Russian airfields, then in one fell swoop destroyed a fleet of Russian bombers. DJI has warehouses of drones across the entire US, all one software update away from being militarized.
-
Computer Vision in Factories for Sustainability at Hampton Court
By
–
Discussing Computer Vision deployed to factories to make them more sustainable, alongside former BP CEO Bernard Looney. At Hampton Court palace in London. Thank you to King Charles and Jennifer Jordan-Saifi for organizing an excellent gathering.
-
Human Learning as Information Compression and Fair Use in AI
By
–
Human learning is information compression. Saying LLMs are “only compressing text” is naive. The brain at birth is a compression of all the experiences of our ancestors related to survival. You can’t copyright “fire is hot, don’t touch it”. It is fair use to learn/compress into
-
Programming Becomes Accessible as English with LLM Tools
By
–
Programming is becoming as easy as English, so sure, most people are capable of it and may just need a small nudge. Picking who to follow is a proxy and people put lots of time into that, so they’ll probably put lots of time into “programming” an algorithm with an LLM