"What will happen to Europe if it keeps ignoring AI?"
Three American labs each (!!) operate more AI compute than all of Europe combined. Today we're launching Europe 2031: a story of what might happen if that doesn't change.
I'm sure I'm not the first to tell you this, but there's really no reliable way to be 100% certain that a statement is written by ChatGPT rather than by a non-native English speaker awkwardly trying to write in fancy English (which is how I wrote my statements)
🚨🚨🚨 VERY excited to share our new paper on how AI can facilitate democratic deliberation, published today in @ScienceMagazine! Together with @mhtessler, @summerfieldlab, and other amazing collaborators at @GoogleDeepMind we've been building the "Habermas Machine"
🚨 Very happy to share new #NeurIPS2022 work with @DeepMind colleagues: "Fine-tuning language models to find agreement among humans with diverse preferences” 🚨
Paper: arxiv.org/pdf/2211.15006…
🧵 (1/n)
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
"Delve" sure seems correlated with increased usage of ChatGPT but that doesn't mean that "Does the text use the word delve" is a sufficiently reliable signal to not invite someone for a PhD interview.
🚨 New paper alert! Excited to share our latest work with the amazing @actuallysoham@jaybaxter and @msaveski introducing Supernotes!
These are LLM-generated @CommunityNotes that synthesize multiple underlying "notes" and are selected by a simulated jury of diverse raters.
🚨🚨 New preprint alert! 🚨🚨
Our new preprint w/ @jaybaxter, @bakkermichiel@msaveski (dream team!) proposes a framework to generate ✨Supernotes✨ - AI-generated notes that synthesize information from several existing @CommunityNotes and are written to foster consensus among
Check out the paper (and the 300 page supplement) for more details at science.org/doi/10.1126/sc…! We also open-source the data, hoping this will be a valuable resource for research in deliberative democracy, social choice, and AI alignment.
Just presented ongoing work on "Fairness in Budget-Constrained Decision Making" (joint with @noriega_campero, @alex_pentland, @krvarshney) at the @kdd_news XAI workshop in sunny Alaska 🌄. Awesome to have a slot right between @_beenkim and Rich Caruana! Video should be up soon!
There are also sooo many tools that can "humanize" your AI written content now (chatgpt.com/g/g-2azCVmXdy-…) and so many different ChatGPT-level models with all different word frequencies. Thinking that you can reliably detect AI content is just a bit silly.
Thanks so much Anca!! Still so happy that you joined GDM and that we now collaborate on the viewpoint pluralism work that was (in part) inspired by the Habermas Machine!
When I joined @GoogleDeepMind last year, I came across this incredible group of people working on deliberative alignment, and managed to convince them to join my team in a quest to account for viewpoint and value pluralism in AI. Their Science paper is on AI-assisted deliberation
The Habermas Machine is a system of fine-tuned LLMs designed to mediate collective human deliberation inspired by Jurgen Habermas' Theory of Communicative Action. Technically, the HM has two three components:
How do we ensure humans can still effectively oversee increasingly powerful AI systems? In our blog, we argue that achieving Human-AI complementarity is an underexplored yet vital piece of this puzzle! And, it’s hard, but we achieved it.
🧵(1/10)
Thanks @medialab for sharing!
If you happen to be at #CHI2019, please come and see brand new PhD graduate Dr @kevinzenghu present VizML (and our other work VizNet) today!