💥 New paper 💥

We discover a form of covert racism in LLMs that is triggered by dialect features alone, with massive harms for affected groups.

For example, GPT-4 is more likely to suggest that defendants be sentenced to death when they speak African American English.

🧵

The way we talk about AI matters. “The model understands how to…” implies much more powerful capabilities than “The model is used to…”
We present AnthroScore, a measure of how much tech is anthropomorphized, i.e talked about in human-like ways. #EACL2024
anthroscore.stanford.edu

🚨 New policy brief: Millions of images are generated each day using text-to-image AI systems. Our latest brief examines how major image generation models encode a wide range of dangerous biases about demographic groups. Read or download here: hai.stanford.edu/policy-brief-d…

New paper (to appear at ACL 2023)! We present Marked Personas, an unsupervised way to measure stereotypes in LLMs for any intersectional identity.
Paper: arxiv.org/pdf/2305.18189…
Joint work with the wonderful Esin Durmus Dan Jurafsky Stanford NLP Group 🧵1/6

Check out our new paper with Dan Jurafsky! What inductive learning biases influence language learning, and how? We pretrain transformer models on different structures and fine-tune on English to test the learning effects of different inductive biases arxiv.org/abs/2304.13060

We have a postdoc position available! Come join our computational social science project on using NLP and speech processing on body-worn camera data to help improve police-community relations! nlp.stanford.edu/job.shtml

📢 Models like #ChatGPT are trained on tons of human feedback. But collecting this costs $$$!

That's why we're releasing the Stanford Human Preferences Dataset (🚢SHP), a collection of 385K *naturally occurring* *collective* human preferences over text.
huggingface.co/datasets/stanf…

Here's the Jan '23 release of draft chapters for Speech and Language Processing. This is just a bug-fixing and chapter-refactoring release, to modernize the structure for teaching/learning. Best wishes from Jim and me for a healthy and happy New Year! web.stanford.edu/~jurafsky/slp3/

Text-to-image generation models (like Stable Diffusion and DALLE) are being used to generate millions of images a day.

We show that these models perpetuate and amplify dangerous stereotypes related to race, gender, crime, poverty, and more (arxiv.org/abs/2211.03759)

A thread🧵

We tend to evaluate multilingual models with downstream tasks. But what if there’s more subtle ways that fluency is affected, like having an English “accent” when learning a second language? Check out our new paper with Kezia Lopez and Dan Jurafsky! arxiv.org/abs/2210.05619

My student Kayo Yin needs your help. Her visa has been unnecessarily delayed, which would prevent her from coming to UC Berkeley to start her studies. Despite bringing all required documents, the Department of State refused to process the visa and it could take months to re-process.

Thrilled to share that I have accepted an offer to join the CS department Stanford University as an assistant professor, starting this Sept. We will continue to work on socially aware and positive #NLP . Very excited to explore new opportunities and collaborations at Stanford 😀

We in Stanford Linguistics have just posted an ad for a tenure-track Assistant Professor (applications due Oct 14, 2022). It's an open area search, and we take a very expansive view of linguistics: facultypositions.stanford.edu/cw/en-us/job/4…

I have a new paper out PNASNews with Ran, Dan Jurafsky, Dallas Card and co-authors. We trace 200,000 Congressional speeches about immigration from 1870-present. Speeches are:
🔸More positive now than ever
🔸But more polarized by party
pnas.org/doi/10.1073/pn…

Excited to share that our paper on the history of immigration discourse in congress has now been published (open access) in PNAS! pnas.org/doi/10.1073/pn…

I will be joining Johns Hopkins JHU Computer Science JHU CLSP as an assistant professor in fall 2023, following a postdoc with Dan Jurafsky! I am very excited and so grateful to all of the people who have supported me, especially my incredible advisor Yulia Tsvetkov