Lmst

I generally dislike puns in article titles but 'careless whisper', about the algorithmic harms of hallucinations in OpenAI's whisper tool, is pretty good https://doi.org/10.1145/3630106.3658996 #facct24

Shocking findings: 38% of hallucinations include explicit harms (violence, inaccurate associations, false assertions of authority) and they are more likely to occur in e.g. aphasic speech — so really this is a bias amplifier #ASR #LanguageTechnology

If you want to know why Mark Zuckerberg irresponsibly doubles down on coopting the "open source" term (and why they're lobbying the EU for 8 million a year): they have a big incentive to count as "open source" under the new EU AI Act — see our scientific paper on open-washing here https://dl.acm.org/doi/10.1145/3630106.3659005

Thanks Zuck for further illustrative material of the #openwashing of #llama3, and for timing it so nicely in the wake of our #FAccT24 paper

Probably the strongest claims to the label open source (if not its content) come from Meta and its Llama 2 and Llama 3 models. The corporate blogposts that introduced Llama made the following claims:

*« “Today, we're introducing the availability of Llama 2, the next generation of our open source large language model.” (Llama 2)

* “Meta has put exploratory research, open source, and collab- oration with academic and industry partners at the heart of our Al efforts for over a decade” (Llama 2)

Comparison of Bloom and Llama showing that only the first can truly claim to be open source; Llama's openness is all marketing and hype.

Meta

Open Source Al Is the Path Forward

o Mtk suekerbers, Founder and CEO

nbd just @andreasliesenfeld keynoting at King's College London on the occasion of OpenUK's Open Manifesto Report Launch, as our #FAccT24 paper on open-washing directly informs policy proposals on AI openness
#proudcollaborator

our paper: https://dl.acm.org/doi/10.1145/3630106.3659005

OpenUK's State of Open: The Open Manifesto Report Launch

ICYMI: a Nature write-up about our #FAccT24 work on open-washing and generative AI https://www.nature.com/articles/d41586-024-02012-5

paper w/ @andreasliesenfeld is here: https://dl.acm.org/doi/10.1145/3630106.3659005 :OpenAccess:

"Rethinking open source generative AI: open-washing and the EU AI Act"

Thank you for sharing! Our #FAccT24 paper "Rethinking open source generative AI: open-washing and the EU AI Act" is available in :OpenAccess: here: https://dl.acm.org/doi/10.1145/3630106.3659005

We write to be read. It's a very special feeling when folks take those words you crafted and whip out the highlighter 🤗

Our #FAccT24 paper on open-washing and #GenAI, out now:
https://dl.acm.org/doi/10.1145/3630106.3659005

Screenshot of a tweet by Gary Marcus saying 🔥🔥🔥🔥, quoting a tweet by #Abebab saying "mic drop" and showing a screenshot with the following quote highlighted:
"When generative AI follows the release-by-blogpost model, it is
reaping the benefits of mimicking scientific communication —including associations of reproducibility and rigour [21 , 23]— without actually doing the work. And when generative AI co-opts the term open source, it is reaping the benefits of libre culture —including associations of transparency and associated freedoms [ 45, 60 ]— without actually contributing to the commons. This is how open-washing works. There is ample evidence that as a communication strategy, open-washing is highly effective. The launches of Llama 2 and Llama 3 were greeted with considerable excitement in mainstream media outlets, almost without exception uncritically echoing the open source claim as a major selling point. For instance, a Wired headline of April 2024 claims, “Meta’s Open Source Llama 3 Is Already Nipping at OpenAI’s Heels”10, and Fortune wrote, “Meta releases its new Llama 3 open-source AI model. Is it enough to keep Meta at the front of the pack?”

Excited to be at @FAccT Conference this week (remotely today, in-person from tomorrow)

Please DM me if you'd like to meet and chat about ML+Law, data protection, political philosophy, labour, environmental justice, 🇵🇸organising against genocide🇵🇸, etc. #FAccT #FAccT24

RT @SeverinEngelma1
📢 New paper at #FAccT24!

We studied pop intro to AI courses on YouTube & developed recommendations for integrating ethics

Joint work with
@madihazc

@ang3linawang

@cfiesler

We find intro to AI courses teach practices known to cause societal harms

Paper https://rb.gy/2ktdik

If you're at #facct24 #FAccT and want to hear me present my linked epistemic power in AI ethics labor paper:

My name isn't on the FAccT schedule for some reason, but my paper is! I'll be presenting it at:

⏰Tuesday, 10:10 AM, in the Azian Room, session IV "RAI work in industry and government"

https://hci.social/@davidthewid/112530701858298695

How can we go consider, but go beyond racial harms in the AI space? How might AI support forms of generative work like creativity and resilience? Find out at our hybrid #FAccT24 panel on Monday at 2:30pm ET, "Black in Generative AI", featuring
Andre Brock, Blu Lewis, and Eric Corbett

#facct24

Client Info