Congrats to Avi!

Also: you can't give a Turing award to eight people, IINM. Something to keep in mind next time you add someone to your paper.

📢 Are you passionate about making a difference through AI research?

Applications are now open for Vector’s Fall 2024 and Winter 2025 Research Internships. Apply today and seize the opportunity to contribute to the future of AI 👇

vectorinstitute.ai/programs/resea…

🏆 We're thrilled to announce the recipient of the 2023 #ACMTuringAward : Avi Wigderson! Wigderson is recognized for his foundational contributions to the theory of computation. Join us in celebrating his incredible achievements! Learn more here: bit.ly/4aGpbiM Institute for Advanced Study (IAS)

Thank you to Jimmy Dick of the Moose Cree First Nation for opening SaTML 2024. He gave us an opportunity to reflect on the land the conference is happening on today.

Somesh Jha is now starting his keynote on watermarking.

Our 12 scaling laws (for LLM knowledge capacity) are out: arxiv.org/abs/2404.05405. Took me 4mos to submit 50,000 jobs; took Meta 1mo for legal review; FAIR sponsored 4,200,000 GPU hrs. Hope this is a new direction to study scaling laws + help practitioners make informed decisions

I like to say that physics research funding may be the biggest hidden subsidy the US government makes to Wall Street.

$200 million towards adoption of AI in sectors like agriculture, health care & clean tech.
$50-million for AI safety institute.
$5.1 million for office of the AI & Data Commissioner.
And $2 billion for computing capabilities & technical infrastructure.
cbc.ca/news/politics/…

Another class of ML papers that collect good review scores largely overclaims the implications and generality of what is actually proven and sweeps the underlying assumptions under the carpet. See below what I ask the referees for those:

This will be my last post until I finish a referee report for a certain distinguished, Renaissance-man-like colleague who must remain nameless lest I out myself as the referee.

The AI Round is tomorrow! We're excited to welcome Roxana Sultan, the Chief Data Officer and VP of Health Vector Institute, as our featured speaker! Interested in how Canada is advancing in the AI & Health sector? This is a presentation you won't want to miss. Looking forward to…

Statistics and ML theory would sound so much cooler if we called everything a power law.

New paper led by Michael Chong on estimating the timing of stillbirths (ante- versus intrapartum), combining a wide range of data sources (which are often only a partial or biased picture) academic.oup.com/jrsssc/advance…

A seriously broken system if the people participating in it don't see the value. We don't have any good alternatives being discussed at large.

New Faculty Q&A: in his research, Assistant Professor Roei Tell studies the limits of resource-efficient computation: What can — or can’t — algorithms achieve when time and other resources are limited? uoft.me/amB

Makes me long for 2014.

When you finish a PhD in computer science, they take you to a special room and explain that you must never use recursion in real life. Its only purpose is to make programming hard for undergrads.

If you need a good reason not to let one company control AI, simply try to have a conversation with, say, ChatGPT about any topic that's touchy.

Yaroslav Bulatov It's called 'investment in the future', a concept that the WSJ and its readers don't seem to understand, given that their attention span is limited to the next quarter.

Objects in mirror may appear nicer than they are in reality.

MSFT stock grants

New hire: can’t code. has 30k twitter followers. $800k over 4 years.

Current eng: finds 500ms backdoor, saves world. $67k retention grant, vests over 5 years.