New postdoc position alert! Excited to announce that our HumAIn project Universitetet i Oslo is recruiting another postdoc. This time in the Cultural Studies dept, on the theme of "Narrative Cultures and Futures of AI”(deadline 31 October): jobbnorge.no/en/available-j…

My (pure) speculation about what OpenAI o1 might be doing [Caveat: I don't know anything more about the internal workings of o1 than the handful of lines about what they are actually doing in that blog post--and on the face of it, it is not more informative than "It uses Python

The Call for Papers ahead of the Oxford-Berlin Colloquium 2024 is now live, with submissions welcome from Early-Career Researchers. Full details on the colloquium and call for papers can be found on the Institute's website:oxford-aiethics.ox.ac.uk/event/oxford-b…

OpenAI’s o1 is impressive What does it mean for AI governance? Some reflections in thread

If I reach 30k followers I might do an Ask Me To Ask Claude Anything. People describe things they want Claude to do but that they think are beyond Claude's current abilities, and I try to prompt Claude to do them.

your little brother 2 hours into the road trip:

Just 2 weeks left to enter into this year's Bernard William's Essay Prize competition! Oxford Undergrads are invited to submit essays in response to: "Is AI a Threat to Human Creativity?" by Monday 30th Sept. 2024 at 5pm (BST). Full details available here: oxford-aiethics.ox.ac.uk/bernard-willia…

Reverse engineering OpenAI’s o1 What productionizing test-time compute shows us about the future of AI. Exploration has landed in language model training. interconnects.ai/p/reverse-engi…

RL has reached another local optimum. It keeps climbing. interconnects.ai/p/reverse-engi…

CALL FOR ABSTRACTS: Artificial Intelligence and Democratic Freedoms Knight First Amendment Institute rebootdemocracy.ai/blog/call-for-…

GPT-o1 and the introduction of 'real' RL into the LLM space changes a lot for AI and security; we were just getting a good mental model of what LLMs could and couldn't do, and now that model needs to be very deeply revisited

never venting to chatgpt ever again

The RIET Lab is hiring! We have open PhD positions with fully funded assistantships in the areas of (1) #AIEthics, #bias, #equity, (2) #AISafety, #Sociotechnical #AIAlignment, and (3) #misinformation, #informationretrieval, #knowledgegraphs. More details: infothreats.cse.uconn.edu/currently-hiri…

This stuff is absolute gold. Thanks La Main de la Mort don’t suppose you'd be interested in presenting your work to a group of philosophers?!

ICYMI: LinkedIn is using your data to train its AI by default. I was opted-in without giving explicit consent; you may be too. The opt-out link is here: linkedin.com/mypreferences/… Steps: Click "Me" on top, Settings and Privacy > Data Privacy > Data for Generative AI Improvement

My lab (MINT Lab) has been awarded a ~USD1m grant from Templeton World Charity Foundation to work on Language Model Agents and Society. Excited to: (1) Anticipate and evaluate societal impacts of LMAs (2) Develop necessary normative foundations and (3) Propose design and regulatory