📢 The online #CriticalAI seminar series returns ✨@ASchjoett@tobias_blanke and I are thrilled to announce the program of the 24–25 series🙌
✨Once again, open to everyone to critically engage with, critique and challenge #AI regimes across different fields of application
Does a USB drive get heavier as you store more files on it?
Nope. Paradoxically (and theoretically), the more you save on a flash drive, the lighter it gets.
USB drives use Flash memory, which means the the ones and zeros of the data are stored on transistors.
When you save data, a binary zero is set by charging the float gate of the transistor, and a binary one is set by removing the charge.
To charge it, we add electrons, and the mass of each electron is 0.00000000000000000000000000091 grams.
This means that an empty USB drive (which mostly holds zeros) weighs more than a full USB drive (which has ones and zeros). Add data, reduce the weight.
Hi all, Hito Steyerl's Keynote for 'Critical AI in the Art Museum' is now up online, visit https://t.co/4I732LrNrf for the recording and supplementary reading.
Don't forget to sign up for our next panel on Tuesday the crisis of the commons in an age of extractive AI.
@yoavartzi@yoavgo@dallascard I was just talking to someone about this idea! Most of the examples I can think of are pre-LLM, maybe it was easier to do model-as-cultural-mirror with more interpretable word embeddings?
I have a bunch of sketchy interface ideas for using language models as epistemic rubber ducks, aka. reflective thinking partners🐥
They've been sitting in Figma for 8 months gathering dust! They all felt too obvious to me to be worth sharing...
https://t.co/p5PNO8gXBc
There’s one existential risk I’m certain LLMs pose and that’s to the credibility of the field of FAccT / Ethical AI if we keep pushing the snake oil narrative about them.
here’s a force-directed knowledge graph interface for @OpenAI’s gpt-4. given a topic, it prompts new questions to ask based on its own generated responses, allowing curiosity-led exploration of a concept.
A look at using GPT3/3.5/4 on library and archives collections. Using the crowdsourced transcribed Susan B. Anthony Papers from @Crowd_LOC as a use case.
Using LLM to manipulate and extract metadata and the type of interfaces/data that makes possible: https://t.co/S4zfiuk4kO
The high pitched AI alarm & glassy eyed celebration & baseless takemaking is exhausting at a soul level & terrifying, painting technologies of centralized corporate power as ~divine.
But you don't talk down a zealot by telling them they're wrong. We need other approaches.
This framing is doing tech monopolies' work for them. Under the guise of alarm, it repeats and extends overhyped claims about system capabilities. Great marketing if you want military contacts, etc. @STS_News has termed this "criti-hype," and here we see it at its least subtle.
who saw LLMs coming?
e.g. decades (or even 5+ years) ago, X said: when machine learning systems have enough compute and data to learn to predict text well, this will be a primary path to near-human-level AI.
Perhaps it was inevitable. What Chomsky did to Skinner has been done to him.
There are serious flaws in this Piantadosi paper, but I suspect it is destined to be a classic — or at the very least, stimulate a lot of conversation.
https://t.co/xQOB9H4L5s
Some highlights.
🧵
Ever wonder how a language model decides what to say next?
Our method, the tuned lens (https://t.co/qeImCg2tMP), can trace an LM’s prediction as it develops from one layer to the next. It's more reliable and applies to more models than prior state-of-the-art. 🧵
What matters about language in the context of this tech is that language and meaning are relational, that communication is a joint activity, and that systems set up to mimic the form of language can provide the illusion that they understand, know things, are reasoning.
>>
Another day, another opinion essay about ChatGPT in the @nytimes. This time, Noam Chomsky and colleagues weigh in on the shortcomings of language models. Unfortunately, this is not the nuanced discussion one could have hoped for. 🧵 1/
https://t.co/nEUNoxUcbY