🚨 NEW PAPER 🚨
@PrathmJ and @Floridi analyze whether Twitter can be used to detect relative reports of polling place issues on U.S. election days.
Available at: doi.org/10.1177/089443932412…
The great problem with LLMs for tons of important public & private sector use cases is hallucination. We won’t solve that anytime soon, but this work from @neilbband and others is exactly where we need to go first. Check it out!
When LLMs are unsure, they either hallucinate or abstain.
Ideally, they should clearly express truthful confidence levels.
Our #ICML2024 work designs an alignment objective to achieve this notion of linguistic calibration in *long-form generations*.
arxiv.org/abs/2404.00474
🧵
Have we got AI news for you! 🤖🗞️Join us at our Roundtable on AI & Elections, 22 May 2-3:30pm BST, where we will discuss the transformative power of AI in shaping electoral processes, focusing on its positive potential & necessary regulatory frameworks!
👉 tickettailor.com/events/tech…
The Computer Says Maybe podcast, hosted by @alixtrot is live! Loved co-hosting the first episode on: AI, our information ecosystem, and the 2024 elections. Surprise, this topic is a lot more nuanced than we're led to believe.
Check it out here: saysmaybe.com/podcast
🎮 Players needed! 🎮 For my @oiioxford PhD, I'm researching how playing Tears of the Kingdom, Skyrim, & Stardew Valley shapes how we think and make meaning. Take a quick survey to share your experiences and indicate if you want to further participate! tinyurl.com/oii-games
New blog from @KeeganMcB and myself: AI is going to transform our elections, making them more personal and less human. And more importantly, AI is going to change what we prioritize in politics, and who gets a say in what that looks like. Read more here:
New blog alert! "How Data and Artificial Intelligence are Actually Transforming American Elections" by @PrathmJ, DPhil Student, and Dr @KeeganMcB, Departmental Research Lecturer in AI, Government, and Policy, both @oiioxford. Read the blog here: oii.ox.ac.uk/news-events/how…
"We ask Twitter to reinstall the old API and to facilitate and expand, not restrict or limit, the quantity of data accessible through it." Here's a petition from myself, @jessRmorley and @Floridi to twitter - sign here: docs.google.com/forms/d/e/1F…oii.ox.ac.uk/news-events/new…
Deeply disappointing news from @TwitterDev - academics currently have free access to 10mil tweets/month, soon to be downgraded to 10k for $100/month. If Twitter is a "public sphere", shouldn't the public have access to its data? Lots of important research will suffer from this
“It's a sad day for our country and our democracy when dominant industries, with assistance from unlimited dark money, get to choose their regulators..with the help of their friends in the Senate, the powerful cable and media companies have done just that"
theverge.com/2023/3/7/236293…
‼️ ‼️ 🎙️🎙️new episode with @oiioxford colleague and @rhodes_trust scholar @PrathmJ this week on political campaigning, technology in US elections and improving access to voting. Such a great chat drawing on Prathm’s political campaign experience and research! Link below 👇🏽👇🏽👇🏽👇🏽
The working paper for my first, first-author paper (alongside @Floridi), "Using Twitter to Detect Polling Place Issues on U.S. Election Days" is now available on SSRN! I'm really excited about this work, and I'll offer a brief summary in the replies:
papers.ssrn.com/sol3/papers.…
Although voting is amongst the most studied fields in political science- there's limited work on the impacts of suppression, because data collection is so difficult. In the past we've had to rely on survey data or observers in a handful of polling places. Now, we can use Twitter.
My PhD has two major goals: examine how digital technologies can decrease barriers to voting and also demonstrate the promise of new methods in political science. I believe that political problem solving is an interdisciplinary task - and this paper is one example of why.