#facct2023 — Public Fediverse posts

Today I #AmReading a fantastic paper that brings an environmental justice lens to algorithmic audits: "Algorithms as Social-Ecological-Technological Systems" by @mozilla Fellow Bogdana Rakova w/ @roeldobbe from this year's @FAccT #AIEthics #AI #JustSustainabilityDesign #FaccT2023

A summary is here: https://montrealethics.ai/algorithms-as-social-ecological-technological-systems-an-environmental-justice-lens-on-algorithmic-audits/ but really, go read the paper...

Next was a nice short talk by Bilel Benbouzid on the sociology of machine learning fairness at @FAccT https://www.youtube.com/watch?v=KmaI64oNWH0 (6/12) #FAccT2023

Next was an incredible slate of talks at @FAccT on the risks of #GenerativeAI and policies to mitigate those risks. Work by Robert Wolfe (sexual objectification in generative #AI), Vinitha Gadiraju (#disability centered perspectives on #LLMs), Irene Solaiman (generative AI model release policies), and @saxon (disparities in text-to-image models across languages) are all highly recommended https://www.youtube.com/watch?v=-n7qzUml6Zs (4/12) #FAccT2023

First was an excellent session at @FAccT on racial #bias in #algorithms. Highlights for me were talks by Robin Brewer (equitable speech technologies for Black older adults) and Teanna Barrett (subjectivity in skin tone annotations). Highly recommend the whole session https://www.youtube.com/watch?v=4J4pLDifsnI (2/12) #FAccT2023 #AI #AIEthics

Next was an interesting talk by Moritz Hardt on technical approaches and implications of algorithmic prediction at @FAccT https://www.youtube.com/watch?v=4ly7zj72zoQ (5/6) #FAccT2023

Next was a nice session at @FAccT on algorithmic #privacy, with a particular focus on technical interrogations of privacy and fairness approaches https://www.youtube.com/watch?v=4pVsm6Jy8GI (3/6) #FAccT2023

Last was a packed session at @FAccT on the limits of #AI explainability. I particularly liked Tim Miller on hypothesis-driven decision support using AI https://www.youtube.com/watch?v=Nl7Hdz40d24 (10/10) #FAccT2023 #XAI

Next was a fabulous talk by Payal Arora on feminist design principles for the future of global work at @FAccT. This is a fabulous tour of the importance of incorporating feminist principles into work, work challenges faced in the global south, and the complex role that technology plays in simultaneously reducing and accentuating those challenges. Highly recommend https://www.youtube.com/watch?v=dkgsmyUE52c (8/10) #FAccT2023 #feminism #work

Next was a rigorous technical session at @FAccT on #bias reduction methods. Brianna Smith's talk on influential fairness was the highlight for me https://www.youtube.com/watch?v=hEypa0tXWM0 (6/10) #FAccT2023

Next was a nice session at @FAccT on recommender systems. I especially liked the work by Rock Yuren Pang on cross-cultural label consistency and by Jessie J. Smith on the challenges of aligning an organization around a single definition of fairness https://www.youtube.com/watch?v=hbJM6vZZmwo (4/10) #FAccT2023

First was a good slate of talks at @FAccT on #algorithms in the wild, with a particularly interesting talk by Matt Franchi on detecting disparities in #police deployments using dashcam data https://www.youtube.com/watch?v=qFgWHdIe6w8 (2/10) #FAccT2023

Next was a nice session at @FAccT analyzing the EU #AIAct. The analysis here is from a variety of perspectives, and is a good place to start if you want to understand the act implications and where AI regulation might be going youtube.com/watch?v=7Dyw7Blh5Cc (6/7) #FAccT2023

First was a great session at @FAccT on values and representation in algorithmic systems. These talks help provide perspective on many of the values that we unthinkingly bake into our systems, and provide a more global perspective on possible ways to think about categories, prediction, and more https://www.youtube.com/watch?v=pY05TteNJ2g (2/7) #AI #AIEthics #FAccT2023

Next was an important panel conversation with data workers at @FAccT with Brook Hansen, Krista Pawloski, and Krystal Kauffman, moderated by @alex. This is a discussion that everyone who deals with data, #AI, #PeopleAnalytics, and #management needs to hear. Highly recommend https://www.youtube.com/watch?v=qZkcZt3qxHQ (7/8) #FAccT2023

Next was an interesting session at @FAccT on #sociology and #MachineLearning. The highlight for me here was the work of Henrietta Lyons on algorithmic decisions and the preference for human reviewhttps://www.youtube.com/watch?v=tFDvd5BVSHg (5/8) #FAccT2023

First was a nice slate of talks at @FAccT on participatory #AI. I particularly liked the talk by @hanlin on how to empower data producers https://www.youtube.com/watch?v=8ihTmVDqHN4 (2/8) #FAccT2023

Next was a fantastic slate of talks at @FAccT on normative #philosophy and #AI. The talks by Borhane Blili-Hamelin on ethical values in #IQ and #ML benchmarks and @sethlazar with a theory of predictive justice. Highly recommend https://www.youtube.com/watch?v=GLKhsrLXvdI (4/5) #FAccT2023 #AIEthics

First was an interesting session at @FAccT on networks and #platforms. I did like the network analysis approaches deployed here https://www.youtube.com/watch?v=FIyf4N5wzXY (2/5) #FAccT2023

Next was an interesting session at @FAccT on algorithmic assessment metrics. These talks get in depth into different approaches to fairness and risk assessment https://www.youtube.com/watch?v=868nTXTq6NQ (7/8) #FAccT2023

Next was an intriguing slate of talks at @FAccT on algorithms and markets. I particularly liked the talks by Daman Deep Singh (gig driver assignment) and Aditya Karan (differential pricing auditing). Highly recommend https://www.youtube.com/watch?v=8WtN3burN-0 (5/8) #FAccT2023 #AIEthics #GigWork

Next was an incredible session at @FAccT on algorithmic power and resistance. Most notable for me were talks by Jacob Metcalf (legal accountability for algorithms) and Dasha Pruss (judicial resistance to recidivism risk assessment). Highly recommend https://www.youtube.com/watch?v=Lt35iArhecg (3/8) #FAccT2023 #AIEthics

Next was a fabulous session on the state of the field of algorithmic fairness, accountability, and transparency at @FAccT. Highlights was work by Eric Corbett (transparency as a non-core ideal), Stevie Bergman (representation in #AI evaluations), Will Hawkins (ethical ambiguity in AI data labeling), and Ali Akbar Septiandri (the WEIRD nature of the conference). Highly recommend https://www.youtube.com/watch?v=NSFiyIHPAYI (8/9) #FAccT2023 #AIEthics

Next was an excellent session on algorithmic #bias and fairness in deployed systems at @FAccT. The focus here is mostly on algorithms deployed by governments, but the challenging issues here are broadly applicable https://www.youtube.com/watch?v=O4MWVMmZNDE (6/9) #FAccT2023 #AIEthics

Next was an incredible session at @FAccT on algorithm trade-offs and impossibility theorems. Work from Andrew Bell, MaryBeth Defrance, Alessandra Calvi, and James Bell are particular standouts, covering the state of the literature on algorithmic fairness and expanding what it can mean and what is possible in this vital space. Highly recommend https://www.youtube.com/watch?v=TamCjQ_tNJo (4/9) #FAccT2023 #AIEthics

First was an interesting session at @FAccT on fairness and diversity, with particularly interesting work by Thibault Laugel (counterfactual explanations for #XAI) and Angie Boggust (saliency cards) https://www.youtube.com/watch?v=beTXuYemKhM&t=4s (2/9) #FAccT2023 #AIEthics #AI

Next was a fantastic collection of talks at @FAccT on algorithmic multiplicity. Amazing talks by @ang3linawang (decision-making algorithm legitimacy), @aaroth (reconciling probability forecasts), Anna Meyer (the impact of unreliable data on predictions), @jamellewd (flexibility and fairness), and @hidden (differential #privacy impact on decisions). Highly recommend https://www.youtube.com/watch?v=bipOsJlSFsk (4/5) #FAccT2023 #AI #AIEthics

First was an excellent slate of talks at @FAccT on model evaluation. Two great talks by Luke Guerdan that deal with labeling issues and measurement error headline this session, but I highly recommend the whole thing https://www.youtube.com/watch?v=gvVBa0etB1Q (2/5) #FAccT2023 #AI #AIEthics

First was a nice slate of talks at @FAccT on algorithmic transparency. I particularly liked @@narrdev's talk on the moral grounding for transparent #AI https://www.youtube.com/watch?v=hMs-EV3r1gc (2/5) #FAccT2023 #XAI

Last was a nice session on #AIEthics in the workplace at @FAccT. The work here is most qualitative but brings some interesting ideas, and I particularly liked the talk by Sonja Mei Wang on future technologies for supporting migrant jobseekers https://www.youtube.com/watch?v=zkdqeSzEv0k (6/6) #FAccT2023

Next was a fantastic session on algorithmic auditing at @FAccT. I highly recommend the whole session, but I particularly liked the talk by @anjalief who shows some surprising results examining racial #bias risk in #NLP tools for child protective services https://www.youtube.com/watch?v=v-H_Z47Vm5E (3/6) #AIEthics #AI #FAccT2023