For all the hype around the “technology-trafficking nexus”, the realities are much messier. New editorial for our series about the intersections of tech and sexual violence, centred around neglected voices of lived experience. From me, @cthibos & @SabraMBoyd for @BeyondSlavery
Sparse Autoencoders act like a microscope for AI internals. They're a powerful tool for interpretability, but training costs limit research
Announcing Gemma Scope: An open suite of SAEs on every layer & sublayer of Gemma 2 2B & 9B! We hope to enable even more ambitious work
I’m super excited to release our 100+ page collaborative agenda - led by @usmananwar391 - on “Foundational Challenges In Assuring Alignment and Safety of LLMs” alongside 35+ co-authors from NLP, ML, and AI Safety communities!
Some highlights below...
After few-shot, one-shot, and zero-shot learning, the next frontier for AI/ML is clear: negative-shot learning
(the machine keeps saying it's a cat though you beg it multiple times to stop)
The fact that most individual neurons are uninterpretable presents a serious roadblock to a mechanistic understanding of language models. We demonstrate a method for decomposing groups of neurons into interpretable features with the potential to move past that roadblock.
Analyzing interview transcripts in qualitative research takes a lot of time.
MyRA is an AI app to help you analyze transcripts and draft reports in minutes.
Here's how to use it:
@AliaiEusebi is presenting our work "The ethics of going deep" at the wonderful first edition of EthiCS 2022 (co-located with @IEEEEUROSP ). Co-authored with @mjvasek and @DrEllaC , full paper available here: https://t.co/LRePFv2WQi
Just out: our new report with @UKAntiSlavery on the war in Ukraine and associated risks of human trafficking and exploitation is now out. Based on a roundtable of over a 100 experts. Co-written with @Aiden_S@UCLCrimeScience https://t.co/1kuzLawKAz
Proud moment: the first paper of a PhD student of mine has been accepted! "The Ethics of Going Deep: Challenges in Machine Learning for Security" will be at EthiCS 2022! Authors @AliaiEusebi@mjvasek@DrEllaC and me, camera-ready pdf available soon!