Thomas Woodside(@Thomas_Woodside) 's Twitter Profileg
Thomas Woodside

@Thomas_Woodside

Policy Analyst @ Center for AI Safety Action Fund. Previously @CSETGeorgetown.

Creator of the beet emoji (forthcoming).

ID:2584689248

linkhttps://www.thomaswoodside.com/ calendar_today23-06-2014 22:43:21

383 Tweets

821 Followers

208 Following

Helen Toner(@hlntnr) 's Twitter Profile Photo

I spoke at the main TED conference in Vancouver a couple of weeks ago! Was a surreal but fun experience.

My talk is up today—it's about how to govern AI, even though we struggle to understand and predict it.

Link below, 🧵 on the main points:

ted.com/talks/helen_to…

account_circle
CSET(@CSETGeorgetown) 's Twitter Profile Photo

⭐️New Blog Post⭐️

AI researchers, journalists, and even members of Congress have referred to claimed 'emergent capabilities' of large language models. Our explainer from Thomas Woodside breaks down the details and the surrounding debate: cset.georgetown.edu/article/emerge…

account_circle
Thomas Woodside(@Thomas_Woodside) 's Twitter Profile Photo

If you find-tune a language model on grade school math (GSM8K) it starts disrespecting content rules.

I recall my 8th grade math teacher also encountered this problem.

More seriously: this is a cool paper.

arxiv.org/abs/2404.01099

If you find-tune a language model on grade school math (GSM8K) it starts disrespecting content rules. I recall my 8th grade math teacher also encountered this problem. More seriously: this is a cool paper. arxiv.org/abs/2404.01099
account_circle