It's almost 2 months since I started at DeepMind, and I can say I feel like I'm finally in the right job. Academia was not the place for me. I love working in teams, and academia has no path for et al's.
A brief update: In a few short weeks I'll be moving to San Francisco to join @OpenAI! I'm both super excited for this change, and very grateful to wonderful colleagues at Google DeepMind for the past 3 years ❤️ Tips on living in SF very much appreciated
🚨 Calling all computational cognitive scientists! 🚨
@OpenAI is hiring. Join us if you're excited to:
🔍 Design & run human/AI behavioral expermts
📊 Collect, analyze, & interpret human data
🐍 Write clean python
🧠 Balance pragmatism with innovation
Job descriptions below!
Our team at DeepMind is hiring! Come and work with me and some really fun people working on human-AI interaction research. Think AI + humans + social science + alignment 🧠👩🔧
I'm at my first ever Neurips this week! Thanks to everyone who came to our poster yesterday on "Fine-tuning language models to find agreement among humans with diverse preferences" with @bakkermichiel, @MartinJChadwick, @mhtessler and others at @DeepMind
#NeurIPS2022
Also disagree on this point. Unless your work is never intended to be read by anyone, why should the code be throwaway? Your code is *a part* of your research and it deserves the same attention to detail as your paper.
So happy my first major piece of work at DeepMind is now published!
We consider a problem at the intersection of cogsci, social science and AI - can AI be used as a force for good, to help groups of people who disagree to find consensus?
Paper: dpmd.ai/3XR3Bm3
🧵…
Could AI help people with a wide range of views find agreement?
Our team fine-tuned a 70 billion parameter language model to generate statements that could bring consensus among groups with diverse opinions: dpmd.ai/3XR3Bm3#NeurIPS2022
Love the pomodoro technique. The regular breaks give me just enough time to make a cup of tea, forget about it, and then rediscover it, cold, 25 min later
as a scientific community can't we all just agree to give one talk only per paper, record it, put it online and stop there? Re-preparing the same talk N times is so inefficient
