Trust worthy machine learning, fairness, alignment, ethics in machine learning...all fundamentally unsolvable problems because it ignores the wider system/world for which the research situates in. Would encourage any young researcher to look into other directions.
3
Incredible colleges and universities can look at this bloodbath and continue to hype up the CS field. Biggest CS faculty recruitment drive I've seen in years.
5
Some much needed sanity amidst seemingly endless amount of publications on reinforcement learning coming out of academia. "Out of 196 peer-reviewed paper from 2018-2024, there is zero case study that reports usage of RL under continuous operation." #useless
8
Probably some academic out there publishing on a topic he or she 100% doesn't understand or know, just so they can call themselves an expert on that topic. This is either done by slapping their names on projects or use AI. This is why research is always doomed.
13
The Bored Academic 💤 retweeted
The world eventually stops caring what you could do and starts judging you solely on what you did. This transition happens quicker than most students expect.
I'm all for social justice but I just don't think it's appropriate to teach it in the middle of a course on industrial PLCs.
25
The Bored Academic 💤 retweeted
现在的大学计算机专业,感觉越来越尴尬了。🤣
学校还在教 Java、Python、数据库、操作系统。
学生课后已经开始用 Claude Code、Codex 做项目了。
我发现现在有两种很极端的学生:
1、一种是连最基本的代码都写不明白,作业全靠 AI,离开 AI 什么都不会。
2、另一种是基础知识学得不错,同时还会用 AI 开发项目,一个人就能干以前好几个人才能干的活。
大学教的这些基础知识倒没什么问题,该学的还是得学。
但现在 AI 写代码都这么猛了,很多学校还在用十几年前那套方式教学生,确实有点跟不上时代了。🤣
Really doubt the quality of textbooks in the pre-AI era. Ran this textbook through ChatGPT and it found 300 errors (with dozens of serious ones) in one chapter of the text. What were the students even learning?
1
39
Observation in academia: no matter how shitty/scammy a college is, the faculty job application procedure is still completely the same; CVs, statements, reference letter. Is this really necessary for "NoWhere International College (NIC)" that's ranked 5400 in QS world ranking?
21
The Bored Academic 💤 retweeted
Meta published a paper that might end the current LLM era.
They proved you don’t need tokenizers to train SOTA models.
Every major AI (ChatGPT, Claude, Llama) reads text in "tokens", chunks of words. This requires massive vocabularies, usually around 100,000 different tokens.
It's how models have always been built.
But researchers at Meta proved this is a massive bottleneck.
They trained a model to read raw "bytes" instead. Only 256 basic units.
The plot twist:
They took a tiny 1-billion parameter model and fed it up to 1 trillion bytes of data.
At first, the standard token-based model was winning in the low-compute regime.
But then something crazy happened.
The token model hit a ceiling. It plateaued. It completely stopped getting smarter.
The byte model just kept climbing.
By converting token probabilities into pure byte-level reasoning (using a new distillation method called "End-Of-Token"), the tiny byte model shattered the performance ceiling.
It matched the standard token model using just 1/6th of the training data.
It crushed storage costs by 80%.
And asymptotically, this tiny 1B-parameter byte model outperforms Llama 3.2 and even the 2-billion parameter Gemma model across the board.
Nobody is talking about the real implication here.
We have spent years forcing AI to read text the way humans do.
It turns out, AI gets much smarter, much faster, when you let it read like a machine.
🚨 ELON MUSK SAYS WE NEED NEW UNIVERSITIES
Elon Musk believes America needs fresh universities that focus on real education instead of wasting money on administration.
He thinks college is overrated for many people. Too often students spend four years and pile up debt without learning useful skills.
What matters most is getting a good education that actually helps you in the real world.
The Bored Academic 💤 retweeted
I'm becoming increasingly skeptical of the entire "AI for Science" narrative.
These days, almost anything can be called AI for Science.
Train a neural network on some scientific data, beat a baseline, improve prediction accuracy by 2%, and suddenly you're accelerating scientific discovery.
But what exactly have we discovered?
A better predictor isn't necessarily a better understanding of nature.
A model that predicts molecular properties more accurately hasn't necessarily taught us anything new about chemistry.
And a model that fits experimental data better hasn't necessarily uncovered the mechanism behind it.
I think we're confusing computational progress with scientific progress.
What's even more frustrating is that much of AI for Science is still evaluated like ordinary machine learning research.
New architecture. Better benchmark. Higher accuracy. Another paper.
But science isn't supposed to be a leaderboard.
The real question should be whether AI helps us formulate new hypotheses, discover previously unknown mechanisms, or design experiments that reveal something genuinely new about the world.
If it does, that's extraordinary.
But if all we're doing is applying increasingly sophisticated models to existing datasets and calling the resulting performance improvements scientific breakthroughs, I'm not particularly impressed.
AI for Science should be judged by the science it produces, not by the AI it uses.
Otherwise, we're just giving machine learning another fashionable name.
The Bored Academic 💤 retweeted
The world needs a high-quality, truly open language model. I wrote about why it’s necessary and why it’s easier to build than you might think.
open.substack.com/pub/argmin…
AI use statements in research paper worth as much as "land statements" made by mega mining corporations.
26
The Bored Academic 💤 retweeted
the real reason people hate AI is bc its forcing an uncomfortable conversation about how much of their work actually matters lmao. when a machine can draft ur emails, summarize ur meetings and make ur presentations in seconds, u start having to explain what exactly the company is paying u for
The Bored Academic 💤 retweeted
The Jacobian tells us how space stretches when we change coordinates.
On a sphere, changing parameters s, t, and h moves a point through three-dimensional space. The Jacobian determinant measures how much a tiny volume expands or shrinks during that transformation.
|det J| = h² / (1 + s² + t²)^(3/2)
It is more than a matrix of derivatives. It shows how the geometry of space changes when we describe it using different coordinates.


















