@tedunderwood.sigmoid.social.ap.brid.gy
I use machine learning to study cultural history at the School of Information Sciences, UIUC. Also #generativeAI, #digitalhumanities, #sciencefiction […] [bridged from https://sigmoid.social/@TedUnderwood on the fediverse by https://fed.brid.gy/ ]
Picture of a home screen where every app icon has been replaced by an appropriate picture of Kermit. For instance, Google calendar is Kermit looking at a calendar, Instagram is Kermit wearing a wig and taking a selfie.
Damon Beres (dlberes on Threads) replaced every icon on his phone’s Home Screen with a picture of Kermit. (GenAI, ofc.) I love the one for X.
A short blog post reflecting on an experience I've had lately: where students show up with a fully-formed solution to a problem I thought was still too difficult for them. What is AI doing to change the way we learn?
#machinelearning #LLMs […]
I see that the Max Planck Institute is hiring two positions in digital humanities, including a Head of #DH. www.mpiwg-berlin.mpg.de/news-events/...
Schmidt Sciences has announced a new funding initiative for work at the intersection of humanities and artificial intelligence. They're also seeking two-page concept papers to help inform the initiative (deadline Aug 23). 🧪 www.schmidtsciences.org/humanities-a...
Abstract: In the U.S., historically, local newspapers drew their content largely from newswires like the Associated Press. Historians argue that newswires played a pivotal role in creating a national identity and shared understanding of the world, but there is no comprehensive archive of the content sent over newswires. We reconstruct such an archive by applying a customized deep learning pipeline to hundreds of terabytes of raw image scans from thousands of local newspapers. The resulting dataset contains 2.7 million unique public domain U.S. newswire articles, written between 1878 and 1977. Locations in these articles are georeferenced, topics are tagged using customized neural topic classification, named entities are recognized, and individuals are disambiguated to Wikipedia using a novel entity disambiguation model. To construct the Newswire dataset, we first recognize newspaper layouts and transcribe around 138 million structured article texts from raw image scans. We then use a
Melissa Dell’s lab continues to produce amazing newspaper datasets. arxiv.org/abs/2406.09490
I just signed this open letter on behalf of the @internetarchive, and hope you will too.
https://www.battleforlibraries.com/
#Books #CDL #Copyright #Libraries
It looks like the new Claude 3.5 Sonnet model is the new best-available LLM - same price as GPT-4o but longer context and appears to be more capable. Some notes here: https://simonwillison.net/2024/Jun/20/claude-35-sonnet/
The https://claude.ai/ UI to it can now build and then render on-demand […]
Too sweet to wait for @nsaphra.bsky.social to post it herself. Can generative (transformer) models achieve better performance than any individual they trained on? Yes, because low-temperature sampling implicitly performs majority voting on diverse data. #machinelearning arxiv.org/abs/2406.11741 🤖
It used to be a philosophical question for science fiction writers, but it's becoming a practical problem for designers. https://tedunderwood.com/2024/06/15/should-artificial-intelligence-be-person-shaped/
It used to be a philosophical question for science fiction writers, but it's becoming a practical problem for designers. https://tedunderwood.com/2024/06/15/should-artificial-intelligence-be-person-shaped/
IMO Charlie Jane Anders
should get royalties on Apple Intelligence because the model of transparent AI — running mostly on mobile, quietly steering the user via reminders without ever quite becoming perceptible as a personality — is just a Caddy from […]
This study found that LLM-generated stories pass significantly fewer Torrance Test for Creative Writing (TTCW) compared to human-written stories, highlighting the gap in creativity. The paper also discusses the potential for LLMs to assist in creative […]
A team centered at McGill and Harvard argues that LLM "hallucination" is better understood as confabulation, and resembles human habits of using narrative for sense-making. #machinelearning 🤖 arxiv.org/abs/2406.04175
Text has been broken into the "tokens" understood by a language model, and each token is assigned a different color. So, for instance "token-ization" is two tokens.
If you're wondering how language models deal with rare words, proper names, and acronyms, this visualizer has you covered. platform.openai.com/tokenizer
From now on I’m checking Night Mode in every dark space just in case there are other phenomena 98% invisible to the naked eye.
Jack Clark’s Import AI newsletter is always fun; this one especially because it explores the shared interest Meta and the CCP have in preventing fine-tuning of their models.
In a “Palantír”-level irony, one of the projects to prevent all further learning is called SOPHON. https://open.substack.com…
If you’d like to make it possible for people to follow you from Bluesky — and you’re in a domain that permits it — all you have to do is follow @bsky.brid.gy.
I think these connections have the potential to strengthen everyone. We need critical mass!
This is a test to see if a message posted in the Fediverse can automatically be forwarded to Bluesky via Bridgyfed:
https://fed.brid.gy/
"Testing, testing, 1 - 2 - 3."
Trying to see if this Bluesky post can make it to the Fediverse on the new bridge.