I teach computer science. They are not, in general, computer whizzes. They are addicted to their phones like everyone else, and that is not the same thing
Alvin Grissom II
Computational Linguist, Associate Professor of Computer Science, Electronic Musician
I couldn't resist.
The doctor actually gave good advice; the article misses the mark. Eating cereal (bad) is not easier than grabbing an apple or some nuts, which have the advantage of being actual food and not chemistry experiments. Corn flakes (for example) send even ostensibly healthy people into postprandial prediabetic glucose ranges, which has serious consequences when done chronically over decades.
Slack scrapes user data by default. Chatting online was commoditized in the 90s.
https://www.securityweek.com/user-outcry-as-slack-scrapes-customer-data-for-ai-model-training/
Looking forward to "verified as human" badges.
https://www.independent.co.uk/tech/internet-bots-web-traffic-imperva-b2339153.html
ML people who view everything through the lens of the mechanisms they're familiar with are making the same mistakes as most medical doctors I've dealt with.
I don't care about the NeurIPS high school track; they can do whatever they want. But I'm glad to see some pushback against the academic pressure cooker environment that programs kids (and future adults) to form their identities around playing these ridiculous games.
It's truly astonishing what a terrible idea this is, but it's an excellent way to de-legitimize your own test.
TIL that a group of ravens is called an "unkindness."
Spock: Perhaps "because it is there" is not sufficient reason for climbing a mountain.
- Star Trek V: The Final Frontier (1989)
https://www.axios.com/2024/03/29/ai-recreate-person-voice-recording-openai
While the sense(s) in which computer scientists use it may have drifted a bit, it's certainly within the same spirit and not intended to evoke comparisons to humans, except (in my case) occasionally to make a point about how these systems *don't* function like people at all.
Point of education: The term "pathological" in a mathematical context predates computer science and has not historically held anthropomorphizing connotations. Computer scientists (including me) typically use the term in this sense. E.g., "pathological case/errors/distribution."
It's unfortunate that fraudulent LLM "reviews" are this common. We all know that peer review quality at ML conferences has tanked in recent years. Kudos to the army of earnest reviewers in the same pool as the ethically challenged ones.
I'm popping in to post this draft since it might be somewhat relevant to recent events. We actually audited the StyleGAN3-r face generation model provided by NVIDIA and found some interesting internal biases that we couldn't explain by the training data alone.
Every web site should have a text-only mode. Please, let’s make this happen.
I’ve been forcing myself to use pandas, and while it definitely has benefits and I’m glad it exists, I’ve found myself:
1. giving up on working around the design philosophy and manipulating CSVs manually; and
2. writing SQL queries over data frames.
I’m curious about opinions on the importance of computational learning theory and support vector machines in a modern into ML class.
I still cover them because the theory of SVMs is beautiful, and I want students to be able to think outside of the box of current trends.
Just realized that I’m academically descended from B.F. Skinner. Ironic.
B.F. Skinner -> William Estes -> David Rumelhart -> Michael Jordan -> David Blei -> Jordan Boyd-Graber -> Me
Haverford is looking for a Visiting Professor of Computer Science. While research is not required, research support is available for this position. Feel free to contact me with questions.
https://www.haverford.edu/provost/news/visiting-assistant-professor-computer-science-1
Client Info
Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst