lowd.bsky.social
CS Prof at the University of Oregon, studying adversarial machine learning, data poisoning, interpretable AI, probabilistic and relational models, and more. Avid unicyclist and occasional singer-songwriter. He/him
390 posts
4,098 followers
487 following
Regular Contributor
Active Commenter
comment in response to
post
I think that's a key question here: do you find AI tools useful or not?
People who don't find them useful think their energy use is a complete waste
People (like myself) who use them every day are much more accepting of their energy costs
comment in response to
post
What I’m getting from this is that days are bad for you and you should try to avoid them.
comment in response to
post
“I’m tired of drinking the Lesser of Two Bleaches. Stop telling me water isn’t bleach. Here’s a study finding that tap water has trace amounts of chlorine in it. Checkmate, you bootlicker for Big Water.”
comment in response to
post
(To be clear, I'm not saying companies are actually good. I don't trust companies, which is part of why I'm in academia. I'd rather talk about science openly than develop it in secret and stop others from using it via secrets/patents/regulations.)
comment in response to
post
I keep hearing how companies are getting rich off AI and also how companies are going broke because AI isn't useful for anything, often from the same people talking about the same companies.
comment in response to
post
Woah. Nice.
comment in response to
post
LLM-created art is a subgenre of found poetry.
comment in response to
post
For image generation, you don’t need every output to be great. If 10% of the outputs are great then it’s fine.
comment in response to
post
My experience has been that the degree of AI hate is much higher on BlueSky, in general. On Twitter, it’s more of a mix.
comment in response to
post
Well, at least not on BlueSky.
comment in response to
post
Maybe not as much as students, but I think I want this!
When students attend my lecture but fail to understand something critical and get stuck in the homework and then can't find the relevant part of the textbook to answer their question... then this is exactly what I'd like my students to have.
comment in response to
post
I haven't yet watched the test-of-time talk, but I remember the original, and he had plenty of hubris 10 years ago! It's always been his style to dream big and make big, enthusiastic claims and not worry too much about caveats. The danger is if too many people put too much weight on his words.
comment in response to
post
When I was first learning guitar and could only play D, G, and A with one strum on each downbeat, I wrote a song.
It’s called “Space Is Big.” Three chords in one pattern, repeated forever.
It’s a really silly song. But it knows it’s silly. And it’s probably the most popular music I’ve ever made.
comment in response to
post
I think I read somewhere that U2 started performing original songs because their cover songs were so terrible.
So, I guess… Whatever you lack in skill, compensate with soul.
comment in response to
post
For example, Louie Louie is a great song.
Also, people still listen to recordings of Bob Dylan singing.
People even listen to Pierce Brosnan sing (Mamma Mia movie soundtrack).
(Not to dismiss your point though — I totally get the frustration of not having the skill to do what you want.)
comment in response to
post
I take comfort in the fact that there’s a lot of great art that doesn’t exhibit high technical skill.
A lot of interesting art comes from the conflict between creative vision and practical constraints, and lack of skill is just one kind of constraint!
comment in response to
post
“The Mitten” is a winter book, not a Christmas book, but it’s so gorgeous that you should at least consider it.
comment in response to
post
The number of TAs in a course is usually an integer, and going from no TA to 1 TA is a big jump, so some of this messiness is inevitable.
You hope to have colleagues that make a good faith effort to recognize excellent contributions to teaching and research even when they don’t align with metrics.
comment in response to
post
Or it can play out in unexpected ways…
Dept: “We provide one TA for every 35 undergrads enrolled!”
You: “Great!”
Dept: “Your class has 30 undergrads and 20 grads! No TA for you! Good luck!”
You: “…wait, what?”
(Not my situation, but stuff like this happens.)
comment in response to
post
I think number of courses is easier to quantify than the other dimensions.
Also, an institution that provides better support might drop that support (run out of qualified TAs, switch to new curriculum in a new programming language with bad textbooks, etc.). Number of courses is a more stable stat.
comment in response to
post
Ugh. Yeah. I’m teaching intermediate data structures to ~85 students with 1 grad TA, and due to the grad student union rules, I’m not allowed to hire undergrads as graders. 😐
comment in response to
post
I’m confused… I find LLMs useful in my daily life. I gave you multiple examples. Am I wrong?
If not, how can you say that we have only established harms?
comment in response to
post
I’ve seen people claim that Google Translate and similar are bad because they replace human translators. There’s a lot of people online really committed to hating anything AI.