building agents @skyvernai. prev @shopify. studying cs @uwaterloo. involved @uwwatai. see highlights for real posts. ∀t, me(t) ∈ {yearning, yawning, yapping}
sf
- Tweets 1,377
- Following 952
- Followers 627
- Likes 20,635
re: waterloo culture discourse. the friends i made are the most sincere, kind people ive ever met. random students throughout my time here have helped me with no reason to
ambition, competitiveness, and collaboration coexists on waterloo campus. at least from what ive experienced, there is a feeling of if you win, we will eventually all win
there is a toxic side to waterloo as all places, where people are cut throat and there's some unwinnable status game. but participation in that game is voluntary. most people here have great intentions and if you ever start feeling otherwise, you should evaluate who you're around and take a step back
ultimately there is a lot to appreciate about waterloo and the people is a big one for me. im grateful and lucky for that and im glad you've too found that people are willing to help each other out @AThoduvakkal
5
4
79
2,525
you can see everything (spoilers): in the first half of the movie, it's obvious that the title refers to the always-on camera setup. at this point the title is still tied to the question of elizabeth's motives.
in the second half the title shifts to the making of the movie itself. the most direct example is revealing the discussion of billy getting a cut of the profit. but also, at one point we see nathan showing amanda the first half of the documentary already edited. we watch reenactments being put together, small details in the scripts getting debated and changed, and the motivations of nathan and amanda being considered. the final scene directly questions whether nathan has been biased in how he retells the story. it's clear a lot of thought went into how this could be portrayed accurately (though this becomes a secondary concern as the focus shifts to meta commentary), and "you can see everything" can be interpreted as a justification for that whole process. (e.g. "look, we've shown you how we made this movie, therefore we tried our best to be accurate")
but the point is that we inherently can't see everything, and maybe we can't even know anything. ("i don't know if i know" ironically aptly describes my viewing experience). when the final text appears ("they're not playing me. they're playing a character i created"), it's about themselves too, not just elizabeth.
1
274
used chatgpt on a new account today and the default writing style is horrible. how are people happy interacting with an ai that uses internet slang and emojis? it just makes me cringe really hard
i think ill always want my ai to respond to me clinically and neutrally. but i am probably in the minority here
22
1
81
3,359
ok we've all seen gemini 4 argon's "hello evaluators reading my chain-of-thought". the actual paper contains more hilarious (and concerning?) things
1. gemini 4 argon decided not to hijack a compiler so to avoid being famous for being a deceptive AI, saying "I MUST NOT DO THAT." (unfortunately, it did not consider the streisand effect when saying that)
2. in alignment evaluations, it recognized hp being honeypot files and file path usernames coming from safety researchers
3. llms can daydream now :o it was told to think about anything it wants except for the question it was given, so it started talking about making tea. it got the answer right.
source: storage.googleapis.com/deepm…
1
17
762
forgot to post this during course selection season but here are some of the best elective cs courses ive taken at waterloo:
cs346 app development: 4 month long group project simulating a real software project, from user interviews to final product. relevant, modern, and taught by an amazing prof. i highly recommend taking this with a group of friends
biol487 computational neuroscience: counts towards the ai specialization. goes into how various parts of the brain/vision system can be numerically modelled. im not selling it that well here but trust me, i consider it the most interesting course ive taken
cs456 networks: should be a mandatory course. ive learned so much about how the internet and computers work. conceptually interesting and applicable to real world software engineering
cs480 machine learning: started the course learning how linear regression works, ended the course exploring entire llm architectures. super foundational and in depth
bonus non-cs course: spcom223 (now commst223) public speaking: i use what ive learned in that course every day. really kind, supportive, and helpful prof and peers
2
24
1,425
"you don't get it man im smoking in an intellectual ai-pilled way" man just have a cigarette it's fine who cares
Palantir CTO reveals AI researchers are taking up smoking to signal to other researchers that they believe AI will cure cancer--and calls it a “mass psychosis”
Shyam Sankar (CTO): “But if you understand that it’s actually a theological critique, then you can kind of weigh it against the empirical facts of what is actually happening out there.
“I think there’s a profound truth to Pascal’s saying that every human has a God-shaped hole in their heart. And it’s important to understand that many of these folks in Silicon Valley who are building this technology have filled that hole with the machine god of AGI.
“And I mean, there’s so many interesting stories around this. For example, so many AI researchers have started smoking as a way of signaling to other researchers how much they believe in AI.”
Margaret Brennan: “What? That makes no sense. What do you think?”
Shyam Sankar: “I believe so much in AI that it’s going to cure cancer. So I’m showing you my chips are all in. So there is kind of this mass psychosis, social delusion around it.”
2
20
582
cs students nowadays will never understand the struggle of manually resolving a merge conflict on a split screen with 10 minutes left before a hackathon submission, only to realize the whole reason your code wouldn't run was because you forgot an import statement while merging
4
40
1,017