Rustic @T1Rustic

take a year out of your life and read these two textbooks cover-to-cover. you will already know more than 90% of people in AI

how to validate your startup idea by @paraschopra 🙌 (link in the comments)

OPENAI IS FINISHED it’s a scam. they’re trying their hardest to retain top position, but they are absolutely unable to in any viable way. they spent a few months saying that their secret o3 model that they didn't want show the public was better than the competition (the public couldn't see it though). with GPT 4.5 they probably released something that's cranked as much as they could, not in a way that makes any kind of financial sense, with the goal of keeping it out there just long enough that they now have a good "existing" model. BUT IF PEOPLE CAN’T USE IT, IT DOESN’T COUNT. Every single company has some shit in the works. Only OpenAI keeps talking about things that doesn't actually exist yet to try to preserve its image as a leader, when it is not leading at all anymore.

I couldn’t agree more. Our graduate students are severely underpaid, and the ongoing federal funding cuts only make their already difficult lives even harder. Imagine how much more productive we could be if students didn’t have to rely on summer internships just to access compute for finishing projects or wait in long queues just to run a small job. Taxpayer money would be far better utilized. Instead, we’re caught in a worsening cycle: funding agencies assume state-of-the-art AI is driven by industry, so they prioritize funding companies for AI research. Meanwhile, academia hires more AI faculty but receives a shrinking budget, leaving labs underfunded. As a result, students are forced to take on TA roles, second jobs, or industry internships —ultimately helping build cutting-edge AI, with all the credit going to the companies.

Wenhu Chen @WenhuChen

2 years ago

Many academia labs (including mine) couldn't even afford a single H100 server. There are much better ways to spend money than 500B mostly on GPUs for one company, which already has plenty of them. That money could lead to extraordinary innovation in academia.

8 11 169 38K 22

@agihippo How I wish I had facilities and food like you. I pack 2days food and just sleep on the office chair and repeat this two days cycle

Our position paper on "Environment Descriptions for Usability and Generalisation in Reinforcement Learning" was recently accepted at ICAART 2025, and is now on arXiv: arxiv.org/abs/2412.16970

Was it necessary to mention someone's race? Afaik, unethical activities have nothing to do with race. The gall!

Jiao Sun @sunjiao123sun_

2 years ago

Mitigating racial bias from LLMs is a lot easier than removing it from humans! Can’t believe this happened at the best AI conference @NeurIPSConf We have ethical reviews for authors, but missed it for invited speakers? 😡

174 777 4K 2.2M 516

90% of European startups lack so much ambition, it’s a tragedy Typical failure cases: focusing way too much on national submarket, product is an undifferentiated generic tooling, pushing marketing before a bold product idea, executing as lifestyle businesses…

Doing a PhD is a huge investment in yourself. You'll know if it's for you. The pursuit of knowledge and the ability to contribute to a field make you powerful. Don't let anyone put that spark out.

Podcasters on political topics are pseudo politicians who might eventually turn politicians. Each one doing extensive research in the problems people face, and some even 'suggest' solutions which are assignments given in school. But I need solutions not problems 🫤

The biggest known collection for speech translation for 13 Indian languages is finally out via BhasaAnuvaad. With 44000 hours of audio and 17M translated text segments, we have pushed the boundary of what was available for Indian languages. This is just the very very beginning and so much more to do for Indian language speech translation, like more and cleaner data, better benchmarks, SOTA models. Have at it, and start churning out your own speech translation models.

AI4Bharat @ai4bharat

2 years ago

🚨🚨 Paper Alert!! 🚨🚨 Introducing BhasaAnuvaad, the largest publicly available speech translation dataset for Indian languages, covering 13 languages and ~44,400 hours. Paper 📄: arxiv.org/abs/2411.04699 Code 💻: github.com/AI4Bharat/Bhas… Dataset 🤗:huggingface.co/collections/ai… 1/N

5 36 230 22K 84

I’m not looking for students this year but I highly recommend the International Max Plank Research School on Intelligent Systems. Come do your PhD at IMPRS-IS!

Intelligent Systems @MPI_IS

2 years ago

Researching #AI #ComputerVision #MachineLearning #Robotics #HCI? Join our elite doctoral program - a partnership with MPI-IS, @Uni_Stuttgart & @uni_tue! Applications accepted until Nov 15, 2024 at imprs.is.mpg.de/application

0 32 36 25K 12

The post we were all waiting for!

Jürgen Schmidhuber @SchmidhuberAI

2 years ago

The #NobelPrizeinPhysics2024 for Hopfield & Hinton rewards plagiarism and incorrect attribution in computer science. It's mostly about Amari's "Hopfield network" and the "Boltzmann Machine." 1. The Lenz-Ising recurrent architecture with neuron-like elements was published in

209 1K 5K 1.2M 3K

@bellandurubcp @east_bengaluru @BlrCityPolice @CPBlr @bellandurutrfps Arrest is good, but what is next? Court will bail him out for money in one week. He may pay for the damage. Case closed. In 2 yrs, he's gonna repeat. I can't imagine how we are all surviving in this country with stupid L&O.

@BlrCityPolice @CPBlr @bellandurutrfps What's happening on Sarjapur Road? A family in car is being attacked by bike brone assailants! Please help! The incident happened at 10:30pm at street 1522, Doddakannelli Junction! The couple just reached police station! #crime #Bengaluru

Fraud by Amazon: I had ordered a Tissot PRX watch on 21st July from @amazonIN. I received the watch on 28th July from the seller Mega Store LLP. I entered its serial number to check its authenticity on Tissot's website. I found that the watch was purchased on 15th Feb 2023.

keyboard_arrow_left Previous keyboard_arrow_right Next

🚨VACANCY🚨: We are looking for a Post-doc candidate in Group Modeling and Explainable Group Recommender Systems! Application deadline: July 14 More info: academictransfer.com/nl/342747/post… #Vacancies #ExplainableAI #RecommenderSystems

Now up on arXiv (and accepted by @ieee_cog): "Towards a Characterisation of Monte-Carlo Tree Search Performance in Different Games". This paper is our first step towards trying to really understand which flavours of MCTS work in which types of games. arxiv.org/abs/2406.09242

keyboard_arrow_left Previous keyboard_arrow_right Next

I’m a competitive person. I like to win. So I find myself as an unlikely advocate of failure on X. I find myself writing about it a lot. When I write about failure, however, I’m really writing about risk. And to embrace risk is to embrace the possibility of both success and failure. Only when you’re really comfortable with the possibility of failure, are you positioned for big success. In the mid-80’s, I was doing my Masters at #Stanford part time and working at a defense-contracting startup. Everything we did was packaged as a “success” even if it wasn’t. If we failed, we wouldn’t get the next contract. So we were allergic to failure and this meant that we were never really, truly, going to succeed. So I wrote a PhD application letter to #Yale that basically said “I want to go to Yale to be able to fail.” Surprisingly, this worked and I got in. I was seeking an intellectually strong environment where people had the confidence to risk and fail. Yale was that place. It was one of the purest and richest intellectual environments that I’ve been in. I didn’t come from an academic family. I didn’t grow up in an Ivy-league world. When I got to Yale, I met people who did have these backgrounds and they had a confidence that I had never seen before. They just believed in their ability to succeed. They were not afraid of failure. Everyone at Yale was pursuing some impossible dream and fully expected that it was possible to achieve it. I went there seeking the opportunity to fail and learned the confidence to attack any problem, no matter how big, with a belief in my ability to solve it. Paradoxically, you can only do this if you are completely unafraid of failure. Older blog post on the topic: perceiving-systems.blog/en/news/the-ro… Medium: medium.com/@black_51980/t…

I often describe #SMPL as a "robot". A robot needs a body with which to interact with world, a perceptual system, a motor system, and the ability to reason and plan. Virtual humans need the same things and SMPL is the body. perceiving-systems.blog/en/news/smpl-a… medium.com/@black_51980/e…