Yonathan Arbel
@ProfArbel
Let's build safe AI! Law prof @ Alabama Contracts, Defamation, Legal NLP, & AI Safety
Why do workers have to wait for 2-4 weeks to be paid, in the same economy where online transactions go quickly and securely? A new draft 𝑃𝑎𝑦𝑑𝑎𝑦-forthcoming @WashULaw-proposes that they shouldn't. Daily, or at least weekly, pay can be a reality. papers.ssrn.com/sol3/papers.cf…
Left-antisemitism is real and pervasive
BREAKING: The largest teachers’ union in America is pushing Holocaust education that erases Jews—referring vaguely to “victims of different faiths.” Imagine teaching about slavery without mentioning Black Americans. Why is this acceptable when it comes to Jews?
the new OpenAI Agent is really good at developing scraping strategies and presumably will power OpenAI's next data squeeze
a side note on the AI action plan: it looks like someone wrote parts to sound like the President before turning to standard policy language
This, but for lawyers. I know many of us live in denial--and this tweet captures beautifully the journey through denial, negotiation, and almost acceptance.
the openai IMO news hit me pretty heavy this weekend i'm still in the acute phase of the impact, i think i consider myself a professional mathematician (a characterization some actual professional mathematicians might take issue with, but my party my rules) and i don't think i…
It is the best model, the most parameters you ever seen, people tell me it gets the best scores. DO NOT believe the lyin' HLE or MMLU that were designed by "AI labs". Thanks for your attention to the matter!
White House going to release a model???
Congrats @openai. I called it! The problem is the other stuff I also think it means
A super interesting initiative at the intersection of generative interpretation and originalism: LLM trained only on literature and legal materials between 1800-1850. It's still not there, but it's a very promising direction
ChatGPT Agent is the first model we classified as "High" capability for biorisk. Some might think that biorisk is not real, and models only provide information that could be found via search. That may have been true in 2024 but is definitely not true today. Based our…
We’ve activated our strongest safeguards for ChatGPT Agent. It’s the first model we’ve classified as High capability in biology & chemistry under our Preparedness Framework. Here’s why that matters–and what we’re doing to keep it safe. 🧵
In the present-future, most initial screening will be done by AI (jobs, grants, pitches...). It strikes me that AI favors stuff written by AI, regardless of merit, because it is more logical from a next-token perspective. That would push the arms race towards greater slop
A new @psalib drop! Download it while your p(doom)<10%!
Pleased to share my new draft with @simondgoldstein. Previously, we've argued that, to reduce x-risk, AGIs should be granted certain basic legal rights. Here, we argue that those same rights are necessary to promote rapid AGI-led economic progress and broad human flourishing.
Pleased to share my new draft with @simondgoldstein. Previously, we've argued that, to reduce x-risk, AGIs should be granted certain basic legal rights. Here, we argue that those same rights are necessary to promote rapid AGI-led economic progress and broad human flourishing.
why can't facts care about MY feelings?
Your regular reminder that tulipmania was a myth. jstor.org/stable/27698044
It will be fascinating to read the history of this era- is this differential success the result of complex organizational dynamics or maybe just the unstoppable personality of a strange, sui generis leader?
I only very partially understand how Grok is possible for a newish lab and how it is impossible for, say, Apple
if the benchmarks are true (and remembering that there are many ways to cook the books), grok 4 is deeply impressive
"the most important for ai safety is [that the model is] truth seeking" lol, wut