PhD student in Computer Science + Communication @ Northwestern
I study and build HCI for science communication
Alum: BITS Pilani, India
she/her
as is now tradition @marianne and I are organizing a small meetup for researchers & practitioners working on HCI, news & democracy at #CSCW2024!
join us for an informal lunch on Nov 12 (Tue) in-person at CSCW. rsvp: https://docs.google.com/forms/u/1/d/e/1FAIpQLSfCdnyalhbNYZAZ43-hPXVgISyMX4XJL_SNY9yVKBTDevJEdQ/viewform?usp=send_form
hi!
@marianne and I are organizing a little meetup for researchers & practitioners working on HCI for #journalism & #news at #CHI2024!
join us for an informal lunch on 14 May (Tue) if you are interested + will be in-person at CHI :)
rsvp here: https://forms.gle/Q8qgUAMF6GWTMw5F7
Our Special Issue on "Artificial Intelligence Controversies" is coming to life, and with a blast: Lucy Suchman on why AI is so remarkably uncontroversial and wildly controversial at the same time. Or: Why AI has accomplished to be taken-granted and push tech-centred discourses on its impact and integration and society.
Imagine someone said fully autonomous self-driving vehicles were inevitable. Why? Well, we've already made "superhuman" vehicles—cars, planes, boats and trains are all capable of moving much faster than humans.
Like, we may still get fully autonomous vehicles, we may not. (I've got the already existing alternative: a city with a good public transport system.)
But "a car is faster than a human" is as irrelevant to the point as how many teraflops a CPU can do is to the likelihood of AGI.
Here's much more on embeddings: what they are, why they're interesting and what you can build with them https://simonwillison.net/2023/Oct/23/embeddings/
Amy's Manifesto: we should take better care of each other, build good stuff, create a sustainable lifestyle for everyone, and scale out love
jazzed to attend #CSCW2023 this year and hang out at the workshop on community-collaborative approaches in HCI!
if you're into participatory design or journalism, let's chat about how to realistically and sustainably build cool stuff with the people who will use it!
Yeah, our software sucks, but the world is just gonna have to deal, you know? Because this kind of software sucks but we're still gonna keep putting out there.
@cscw I love you thanks for always being so kind to me and my work, Costa Rica here I come!!!! (why is this more than a year away gah???)
@jordant down for me too, yeah!
@natematias @cyberlyra some protection for public interest work that also leverages machine learning etc. sounds crucial. although, I am biased: as a researcher who studies journalism and builds tools for it using simple stats stuff and machine learning over news data (not for profit), I am kind of unsure what this means for some of the work I’ve done. would be curious to see how the NYT starts enforcing this and who for to begin with.
Something I discovered today is if you Google for "real estate novelist" you get a "featured snippet" sourced from an AI-written webpage claiming a "real estate novelist" is a novelist who writes about real estate. However, there is a trick here: the AI page is not conventional SEO but rather someone who as a joke asked an AI to write about "real estate novelists". So this is an intentional attempt to poison Google results, and therefore simultaneously less funny, more funny, and less dystopian.
People keep saying "AI" but meaning "web scraping".
"Rather than meet societal obligations to invest in education and physical and mental care, AI’s advocates risk creating a two-tier system where artificial facsimiles will be deemed good enough for those without means—but the well-to-do will hire actual humans."
https://techpolicy.press/ai-hurts-consumers-and-workers-and-isnt-intelligent/
New from me & @alex on Tech Policy Press
"AI is going to take your job!"
Like hell it is. Try:
"A super-rich capitalist has invested billions of dollars into replacing human creative work with a shitty computer program, so he can pocket more of the profits."
When you stop intentionally obscuring who is doing things and who is benefiting from them, the world starts looking very different indeed.
Pro tip: If you define the values at the start of the project, and develop aligned to those values -- including pretraining data curation -- you don't have to "align" values post-hoc. Prove me wrong.
What happens when value-laden design artifacts are put into practice? Do they exercise their commitment to stated design values? @sayamindu and I now have a #DIS2023 paper that attempts to answer these questions. Also, we got a best paper award 🎉 Paper presentation tomorrow (Friday) at 11 am EDT.
📃 Preprint Thread! 📃
I'm so excited to share "Misclassification in Automated Content Analysis Causes Bias in Regression. Can We Fix It? Yes We Can!" by myself, @hase_va, and @chainsawriot We demonstrate a widespread problem and propose how to solve it.
Computational social scientists, in our discipline of Communication and beyond, often use machine learning (ML) classifiers to automate content analysis by building variables for regression analysis.
