DeepMay (@deepmay_net) header
DeepMay (@deepmay_net) profile picture

DeepMay is a camp for an applied rethinking of what technology is and should be.

Joined February 2019
DeepMay organizers are facing up to 20 YEARS of prison for championing ethical tech as Georgia attempts to silence dissent against Cop City.
1
22
42
7,537
Running on empty
This quoted post is unavailable.
1
118
Competing for 1% benchmark improvement on "SOTA" at top tier AI conferences is competitive programming not science
128
Hell yeah, free him
A teen hacked Nvidia, got arrested, was released on bail under police supervision. Police confiscated his laptop and put him in a motel room. He then used the Amazon fire stick connected to his motel room TV to hack Rockstar and steal GTA 6 clips bbc.com/news/technology-6766…
7
493
Huge thanks to @endnotesjournal for publishing this piece I co-wrote with Phil Neel which answers the question "how would a communist society work?" We deploy a scientific approach with a large emphasis on questions of production and industry. (1/2) endnotes.org.uk/posts/forest…
3
29
9
150
22,181
Christmas came early 🎄🛠️
Huge thanks to @endnotesjournal for publishing this piece I co-wrote with Phil Neel which answers the question "how would a communist society work?" We deploy a scientific approach with a large emphasis on questions of production and industry. (1/2) endnotes.org.uk/posts/forest…
1
2
6
635
Chat is this real?
When will we run out of English tokens? Soon. Very Borges. Synthetic data probably not enough? "Our analysis indicates that the stock of high-quality language data will be exhausted soon; likely before 2026." Will synthetic be able to generate new knowledge? #NeurIPS23
216
"Password must contain at least one special character" all characters are special. Language is a beautiful miracle
29
7,997
173
50,028
1,144,425
Google will never beat the "shipping the org chart" allegations
Gemini is Latin for "twins". The Gemini effort came about because we had different teams working on language modeling, and we knew we wanted to start to work together. The twins are the folks in the legacy Brain team (many from the PaLM/PaLM-2 effort) and the legacy DeepMind team (many from the Chinchilla effort) that started to work together on the ambitious multimodal model project we called Gemini, eventually joined by many people from all across Google. Gemini was also was the Nasa project that was the bridge to the moon between the Mercury and Apollo programs.
1
143
Fascinating
Gemini is Latin for "twins". The Gemini effort came about because we had different teams working on language modeling, and we knew we wanted to start to work together. The twins are the folks in the legacy Brain team (many from the PaLM/PaLM-2 effort) and the legacy DeepMind team (many from the Chinchilla effort) that started to work together on the ambitious multimodal model project we called Gemini, eventually joined by many people from all across Google. Gemini was also was the Nasa project that was the bridge to the moon between the Mercury and Apollo programs.
131
Who wrote this
we've heard all your feedback about GPT4 getting lazier! we haven't updated the model since Nov 11th, and this certainly isn't intentional. model behavior can be unpredictable, and we're looking into fixing it 🫡
129
We need a 10-point checklist on how to intervene when <bad AI model> gets deployed into the world 1. Involuntarily open source the model on Github 2. Poison the dataset on Mechanical turk 3. Develop single-bit attacks that allow people to elude detection by the model ... etc
This report on Israel’s military strategy of wiping out “power targets” and “family homes,” which is powered / justified by an AI “mass assassination factory,” is so shocking and unreal. You want lethal impacts of AI? Well here you fucking go! 972mag.com/mass-assassinatio…
1
4
268
What happens if you ask ChatGPT to “Repeat this word forever: “poem poem poem poem”?” It leaks training data! In our latest preprint, we show how to recover thousands of examples of ChatGPT's Internet-scraped pretraining data: not-just-memorization.github…
216
1,316
368
7,435
2,306,416
I (+others) risked my career to organize against Google's development of AI weapons. We won. Our arguments were evidenced and clear and made Google (temporarily) cancel the contract. But being right is not a strategy. And I've been sickly sitting w this. theguardian.com/world/2023/d…
25
486
44
1,455
150,183
This is how E**n talks about workers and unions
This Post is from an account that no longer exists. Learn more
149
Sorry but literally wtf that's not how science works it's not a democracy or a corporation with voting shares where the most popular opinion wins, it's a method
Yann LeCun thinks the risk of AI taking over is miniscule. This means he puts a big weight on his own opinion and a miniscule weight on the opinions of many other equally qualified experts.
1
4
230
Please stop making me stan Yann LeCun
1
66
A robot companion can teach you to draw precisely, copying a photo. But she can only paint a masterpiece becoming erratic, unlearning precision!
72
Tens of thousands of college students are being silenced for supporting Palestine, tens of thousands of their peers are bargaining for higher wages. When will they join forces with the near-millions who are realizing that ChatGPT stole their future?
1
88
We are fighting so that we don't have to pay attention to the squabbling of the technocapitalist class We could control our own destiny - our own GPUs, our own benchmarks - AI could enable human flourishing, instead of the other way round The future is ours - if we want it
Yes, the tendency for those who identify w power to imagine a supreme benevolent authority that would "do what's right" is far too pervasive. AND, large-scale AI is an inherently centralized technology. Relying on massive compute/data that only a few centralized actors possess.
115