Intervention from @e_farries and @abebab on the Gardai's frothing appetite for what is really just bad technology and over-surveillance.

(Paywalled)

https://www.independent.ie/opinion/comment/elizabeth-farries-and-abeba-birhane-giving-facial-recognition-tech-to-gardai-will-create-a-policing-nightmare/a1003081427.html

We also find that most published audit work focuses products/models/algorithms — so we echo calls from others to expand audit work to include the entire ecosystem of affected stakeholders.

I learned a lot from this study — some of the most impactful audit work is done academia (e.g. by journalists or regulators) often using very different methods.

Excited to share a new paper with @abebab, Victor Ojewale, Briana Vecchione & Deb Raji

We surveyed 300+ AI audit studies from academia, civil society, govt etc. to understand what work is being done + how it relates to impact & accountability.

https://arxiv.org/abs/2401.14462

“Japan: Foreign-born residents sue government for alleged racial profiling”

https://www.bbc.com/news/world-asia-68137532

@mekkaokereke

The only kneeling I want to see before a sportsball game is a Black Lives Matter protest.

Drafted a banging 1000-word piece that links the Post Office scandal to the future harms of AI. It's with the University's press office but if you know any outlets or contacts that might be interested, please let me know.

"A group of hackers gained access to the backend of an AI chatbot that fast food franchises use to help automate hiring. The researchers were seemingly able to accept or reject specific job applicants, and had access to a wealth of sensitive information on applicants, the fast food franchises, and the AI company itself, called Chattr."
https://www.404media.co/hackers-break-into-hiring-ai-chat-bot-chattr/

So mastodon doesn't do quoted posts. I think I understand the goal of avoiding that feature. People have decided that "dunking" is the overriding usage of quoting and that we should avoid the inherent toxicity.

But quoted posts are also the foundation of true discourse in a medium like this. It is the primary way to start *new* conversation that is an offshoot of an existing conversation rather than a direct response.

I'm finding it hard to do that on here. And the result is I just don't post.

👴🏻How can a person be expected to defend themselves against accusations of plagiarism with only 1 day notice?

👩🏿How many days advanced notice did Claudine Gay get before her story dropped?

Y'all aren't serious. You want different rules for you vs attacking Black women.

But no.

I used @abebab et al’s papers in my teaching in an *Education* dept back in Spring 2022. For techies not to have cited it in late 2023 seems… myopic, to put it kindly! I hope they will do the right thing & update their paper ASAP.

Another paper by Abeba Birhane and Vinay Uday Prabhu had already resulted in the Tiny Imagenet dataset being taken down.

https://ieeexplore.ieee.org/abstract/document/9423393

404 Media reports that "Largest Dataset Powering AI Images Removed After Discovery of Child Sexual Abuse Material" 🧵

However, in 2021, a preprint by @abebab, Vinay Uday Prabhu & Emmanuel Kahembwe found a number issues in the dataset including " troublesome and explicit images and text pairs of rape, pornography, malign stereotypes, racist and ethnic slurs, and other extremely problematic content."

The preprint can be found here: https://arxiv.org/abs/2110.01963

https://www.404media.co/laion-datasets-removed-stanford-csam-child-abuse/

@hipsterelectron i wouldn't be surprised tho if this were true — but another disappointing thing about the @404mediaco coverage is that it at no point mentions the serious and widely known auditing work of LAION by @abebab and colleagues (e.g. recent Neurips paper https://openreview.net/forum?id=6URyQ9QhYv ); a clear form of bropropriation, only white men and Stanford folks get to shine in this story

'about if decolonial AI is at all possible '

Chapter 11: It’s incomprehensible: on machine learning and decoloniality
Abeba Birhane
and Zeerak Talat

@abebab

https://www.elgaronline.com/edcollchap/book/9781803928562/book-part-9781803928562-16.xml

@axbom you're too kind, thank you!

Our work on AI for medium-range weather forecasts is published in @ScienceMagazine today. 🎉 A lot more for us to do as a community in this area; we hope these advances will support the vital weather-dependent decision-making that happens each day. https://www.science.org/doi/10.1126/science.adi2336

@samhforbes big thanks to @abebab @RealSexyCyborg @Iris @mariekewoe @KirstieJane and others off this platform... for their feedback, discussions, support 🥰

@algorithmwatch There's a bunch of names in there I trust, at least (e.g. @abebab, @vdignum, Nighat Dad). Agree that more representation from civil society would be helpful, however. Are there any particular people you're thinking of?