💻❤️ #CSforALL i work at Google Research
Massachusetts
Joined January 2013
- Tweets3K
- Following3.4K
- Followers1.6K
- Likes5.9K
Kevin Robinson retweeted
Let's goooo!!! The second of two papers on AI education is coming up in a bit. As an AI educator focused on inclusion and co-generative pedagogy, I'm *really* excited for this talk on exclusionary pedagogy. Will tweet some take-aways in this thread:
Stream 2 ⬇️⬇️⬇️
1)Avoiding Disparity Amplification under Different Worldviews by Samuel Yeom & Michael Carl Tschantz
2)“You Can’t Sit With Us”: Exclusionary Pedagogy in AI Ethics Education by Inioluwa Deborah Raji, Morgan Klaus Scheuerman, & Razvan Amironesei #facct21 #facct2021
Kevin Robinson retweeted
so, while the technical work of algorithmic transparency is important, it should be part of a broader perspective on how these systems are engineered, produced, and used. having a nice technical model of a dog's vision system won't alone tell me if it will bite me. (6/7)
Kevin Robinson retweeted
My @FAccTConference @DiverseInAI talk Fairness Metrics & Non-Discrimination Law: Can Fairness be Legally Automated? is online ft my paper Bias Preservation tinyurl.com/5fefced4 & Why Fairness Cannot Be Automated tinyurl.com/jytrxfmc @oiioxford @turinginst @CompSciOxford
Kevin Robinson retweeted
A year or so ago while working on #datacards, I told @ZaldivarPhD "transparent documentation seems like an opaque deal. We need to make it more human." Today, we are stoked to open-source 1/2 of the #datacardsplaybook & present it at #facct2021
#facct21
Pair-code.github.io/datacard… – at Toronto, Ontario
Kevin Robinson retweeted
I need a name for that threshold when research findings about a category of people does more harm than good, when evidence becomes stereotype that hurts the exceptions. Favored research methods elide human diversity & complexity. #research #inclusion #DataScience
Kevin Robinson retweeted
What are the responsibilities of researching AI in the Global South? Many common to all fieldwork: How can research be productive not extractive? Awesome #FAccT2021 tutorial "AI on the Ground" with incredible scholars @VidushiMarda @so_radhikal @tetisheri + Ranjit Singh!
Kevin Robinson retweeted
I'm busy running this thing, but looks like we have some useful live-tweeting of our tutorial on teaching ethics in the tech classroom! 😍 #FAccT2021
Moments away from an *incredibly* important panel on ethical thinking in the tech classroom, with @cfiesler, @s010n, @achaintreau, @_jessiejsmith_, and @michaelzimmer here at #FAccT21. Looking forward to the learnings and provocations.
Kevin Robinson retweeted
Excited to speak today
@FAccTConference 12:00-12:45 UTC on Fairness Metrics & Non-Discrimination Law: Can Fairness be Legally Automated? 2021.facctconference.org/age… talk will also be recorded @oiioxford @CompSciOxford @BKCHarvard @turinginst @oxsocsci @Ox_CyberSec @OxfordLawFac
Kevin Robinson retweeted
Fairness research in ML remains largely rooted in Western concerns: the injustices, datasets, measurement scales, & legal tenets used. These infrastructures, values, & legal systems cannot be naively generalized to non-Western countries -- @autopoietic arxiv.org/abs/2012.03659
![1 Introduction Fairness research in Machine Learning (ML) has seen rapid growth in recent years; however, it remains largely rooted in Western concerns and histories: the injustices they focus on (e.g., along race and gender), the datasets they study (e.g., ImageNet), the measurement scales they use (e.g., Fitzpatrick), and the legal tenets they draw from (e.g., equal opportunity). This implicitly Western take is troublingly becoming a universal ethical framework for ML, e.g., AI strategies from India [1], Tunisia [51], and Mexico [32] all derive from this work, but fails to account for the several assumptions conventional algorithmic fairness makes about the availability and efficacy of the surrounding institutions and infrastructures. However, these infrastructures, values, and legal systems cannot be naively generalised to various non-Western countries.](https://pbs.twimg.com/media/EpT31_KUYAA7tPi.jpg?name=orig)
ALT 1 Introduction Fairness research in Machine Learning (ML) has seen rapid growth in recent years; however, it remains largely rooted in Western concerns and histories: the injustices they focus on (e.g., along race and gender), the datasets they study (e.g., ImageNet), the measurement scales they use (e.g., Fitzpatrick), and the legal tenets they draw from (e.g., equal opportunity). This implicitly Western take is troublingly becoming a universal ethical framework for ML, e.g., AI strategies from India [1], Tunisia [51], and Mexico [32] all derive from this work, but fails to account for the several assumptions conventional algorithmic fairness makes about the availability and efficacy of the surrounding institutions and infrastructures. However, these infrastructures, values, and legal systems cannot be naively generalised to various non-Western countries.
Kevin Robinson retweeted
We need to upgrade our dataset culture in ML. If we ever hope to make our models work, data needs to be more than an afterthought - its the engine driving the field & we need to treat it as such!
Inspired by @timnitGebru & others who continue asking questions on ML's data use.
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Kevin Robinson retweeted
Continually grateful to @mmitchell_ai and @timnitGebru for creating a team that supported and fostered this kind of work. Also huge credit to them both for their seminal research in this area, which has paved the way for this work.
When I read this paper I thought to myself, yes, this is correct, how have we not already been talking about data like this? This is just one example of the excellent, impactful work that's come out of the Ethical AI team that was co-led by @mmitchell_ai and @timnitGebru
Kevin Robinson retweeted
Submit a workshop to FAccT!
In related news, the call for CRAFT workshops at @FAccTConference 2021 is live! We're looking for workshops that creatively engage with and address critiques of the FAccT field. Please share far and wide.
facctconference.org/2021/cfw…
Kevin Robinson retweeted
In related news, the call for CRAFT workshops at @FAccTConference 2021 is live! We're looking for workshops that creatively engage with and address critiques of the FAccT field. Please share far and wide.
facctconference.org/2021/cfw…
Kevin Robinson retweeted
Re-reading this bit right now in @sulin_blodgett's awesome PhD thesis. Seems relevant
.@sulin_blodgett's NLP bias review is great for many reasons. But just this quote on Siri's non-support for African-American English, from @ruha9 - wow
Kevin Robinson retweeted
I'm looking for funded opportunities for interpretability + #NLProc research for summer 2021 (internships + other). If you know of anything, please pass along!
Kevin Robinson retweeted
"...this kind of work is the definition of what it means for computer science research to be impactful in the field."
starting at 1:59 in m.twitch.tv/videos/771560387
Kevin Robinson retweeted
hello! this paper is amazing, both in its ambitious and expansive aspirations for ML, and at the same time its thoughtful humility about the rest of the world outside ML :) thanks for sharing such awesome work!
Kevin Robinson retweeted
I learned so much from @timnitGebru at the Ethical AI team! She is an incredible human and an amazing researcher. She made an environment that we were enjoying every moment of working there. Besides her research, she works on educating the community! I just can’t believe this!
Kevin Robinson retweeted
ALSO, we will also discuss @databoydg's paper on Combatting Anti-Blackness in the AI community: devinguillory.com/files/AI_A…. Bring art/poetry/music as suggested by @shakir_za too.
Join us tomorrow (Saturday) from 17:00-19:00 GMT for an informal @black_in_ai social at @icmlconf where we'll have a meet and greet and be in conversation with @shakir_za and @png_marie on Decolonizing AI (read their paper arxiv.org/abs/2007.04068).
