Daniel Hașegan (@dhasemusic) header
Daniel Hașegan (@dhasemusic) profile picture

🧠 💻 👨‍🔬🎸🇷🇴 Grad student at @nyulangone Brain dynamics and perception @biyuhe lab. Former @Twitter @BraidHealth . Opinions are my own; RT≠endorse

New York City
Joined February 2012
Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)Photo by Daniel Hașegan (@dhasemusic)
10/ #Result 5: Can scaling data, models, or compute fundementaly solve this? We don't think so! #OpenAI's #o1-series is performing better but still suffers from slight performance variations. #o1_preview shows significant improvements, but...
4
15
4
194
42,854
“I'd also like to acknowledge my students (…) they've gone on to do many great things. I'm particularly proud of the fact that one of my students fired Sam Altman.“ 😳🫡
89
652
228
7,654
835,799
Demis Hassabis won the nobel prize! Now we know that training in cognitive neuroscience does pay off!
2
102
Daniel Hașegan retweeted
Everyone's talking about Sakana's AI scientist. But no-one's answering the big question: is its output good? I spent hours reading its generated papers and research logs. Read on to find out teapawt.net/SakanaAILabs/status/18…
Introducing The AI Scientist: The world’s first AI system for automating scientific research and open-ended discovery! sakana.ai/ai-scientist/ From ideation, writing code, running experiments and summarizing results, to writing entire papers and conducting peer-review, The AI Scientist opens a new era of AI-driven scientific research and accelerated discovery. Here are 4 example Machine Learning research papers generated by The AI Scientist. We published our report, The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery, and open-sourced our project! Paper: arxiv.org/abs/2408.06292 GitHub: github.com/SakanaAI/AI-Scien… Our system leverages LLMs to propose and implement new research directions. Here, we first apply The AI Scientist to conduct Machine Learning research. Crucially, our system is capable of executing the entire ML research lifecycle: from inventing research ideas and experiments, writing code, to executing experiments on GPUs and gathering results. It can also write an entire scientific paper, explaining, visualizing and contextualizing the results. Furthermore, while an LLM author writes entire research papers, another LLM reviewer critiques resulting manuscripts to provide feedback to improve the work, and also to select the most promising ideas to further develop in the next iteration cycle, leading to continual, open-ended discoveries, thus emulating the human scientific community. As a proof of concept, our system produced papers with novel contributions in ML research domains such language modeling, Diffusion and Grokking. We (@_chris_lu_, @RobertTLange, @hardmaru) proudly collaborated with the @UniOfOxford (@j_foerst, @FLAIR_Ox) and @UBC (@cong_ml, @jeffclune) on this exciting project.
29
167
34
1,134
472,866
Daniel Hașegan retweeted
Announcing new paper from our lab published yesterday in @NatureComms , led by the amazing @YuanhaoWu. "Network mechanisms of ongoing brain activity’s influence on conscious visual perception". Summary picture below. 1/5 nature.com/articles/s41467-0…
2
49
197
25,334
Daniel Hașegan retweeted
プロダクトから、果物・野菜へ。きのこも美しい。食べ物はみんな美しいと思う。(5/7)
31
2,624
125
17,430
1,074,235
Black mist appears to approach.
18
69
9
406
36,419
Daniel Hașegan retweeted
I had this exchange with Elon here on Twitter over a year ago and we are still having the same conversation so I guess…let’s do this. Fortune 500 companies aren’t overly moral actors. They make decisions based on whether they think they will make more or less money. Advertisers are not leaving Twitter because they are trying to make a statement or achieve some goal (which would be a boycott). They are leaving Twitter because they aren’t sure whether advertising on the platform is delivering negative or positive value, and why spend a bunch of money doing something that might actually be /hurting/ you. The reason it might be hurting them is fuzzier. But, in my opinion, it’s not mostly to do with their ads showing up next to bigoted content. I think it’s that this platform is increasingly all about one guy in a lot of people’s minds. That one guy is both the most influential power use of the platform, and the owner of the platform. And that guy has trained everyone to expect that he is going to say increasingly inflammatory and unhinged stuff. That means employees and customers of those companies (who /are/ moral actors) don’t get good vibes from the company they like advertising with a company they don’t like. This is a symptom of the culture war that Elon Musk desperately wants to inflame because he is stuck inside of a really tight “you are always right” feedback loop of fans and friends who feel incapable of questioning him. He has attained a kind of moral certainty that he is fighting a war against an existential threat to humanity. That inevitably results in either an intervention of good friends or a complete crash (often times a violent one). But that’s when the person isn’t the richest person in the world. I don’t know what happens when it’s the richest person in the world, but I’m trying not to watch it like it’s a train wreck, because, just like with train wrecks, real people are getting hurt. That moral certainty is allowing him to do things that are very clearly bad for the economics of the platform (but good for keeping people talking about Elon Musk, which (as a person who has fallen into that trap before) is important to keep an eye on as an underlying motivation for all of this behavior.) It is not at all surprising that large advertisers, who advertise specifically to ensure their brands are perceived positively, are leaving (at least for the moment) a platform whose biggest brand ambassador is tweeting Great Replacement theories and Pizzagate memes. It is not an attempt to change Twitter or Elon’s behavior. It is an effort to not lose money.
613
2,553
210
35,433
3,831,836
Daniel Hașegan retweeted
Excited to announce the Center for Neural Science (CNS) at NYU is searching for a new tenure-track faculty member — applications are due January 1 — Please RT! apply.interfolio.com/136781
86
5
134
48,620
Test 1 2, is this still working?
2
242
Daniel Hașegan retweeted
My article summarizing our recent workshop on consciousness at the NIH (June 2023) is now out in @NeuroCellPress! 50 d free-access link: authors.elsevier.com/a/1hxjg… This article presents a vision for consciousness research ahead that is more empirically grounded & broader in scope.
1
34
1
99
21,535
Wild—just by flickering some colors you can make your brain hallucinate a triangle that doesn’t actually exist, moving in any direction you want, even though it doesn’t actually move. – at Manhattan, NY
動いてません
2
2
9
1,912
As a member of ASSC and part of two TWCF-funded initiatives, I signed the IIT letter. I have some further thoughts on why I signed that I think are worth reading (apologies that this is long!) and worthy of further discussion: paclab.psych.ufl.edu/conscio…
4
32
6
79
45,635
I read about IIT in 2018 when I was preparing for some career changes. As an outsider, I thought that IIT is a leading theory of consciousness but I was not satisfied after reading Koch's and Tononi's books. Hence I decided it's not for me and I could use my skills elsewhere 1/n
2
2
19
8,007
Without neuroscience background, it was hard to gauge the science of C while IIT was portrayed as the leading theory. Nowadays, I find myself in the consciousness field anyway, thanks to outreach from scientists like @meganakpeters . 2/n
2
1
8
1,232
While I am not fully convinced of labeling IIT as pseudoscience, the leading-theory label put me back a few years where I could have done research. 3/3
1
8
1,954
Daniel Hașegan retweeted
Saw this tweet a few days ago and have thought about it multiple times since
5
31
239
13,094