📢 Are you fed up with the binary discussions between AI existential risks and AI social/ethical risks?
Then read my new manuscript
👇🏻. Feedback very welcome.
@nytimes
THIS DID NOT HAPPEN; someone said something. The article starts with a hazardous use of the words: "Iran has abolished the morality police ". First,
@vivianhyee
, the reporter, apparently has no idea how layered and multi-faceted "Iran" is. Whom does this statement refer? 1/2
📢 Searching for academic/non-academic researchers who work on non-Western AI ethics. Reason? Inclusion in a couple of forthcoming AI ethics events to increase the diversity of perspectives. Please refer me to names either in the comment or in private message. Thank you 🙏
Two professional updates: I'm excited to join
@CarnegieMellon
as an asst prof later in 24. I'm also honored to be selected as a
@SchmidtFutures
AI2050 Early Career Fellow, alongside an exceptional cohort; commencing my fellowship after starting at CMU!
📢 Call for Papers
"AI Ethics in the Generative AI Era" 🌐
Submit your work on the societal and ethical implications of the GenAI, and contribute to a rigorous, interdisciplinary analysis of GenAI's opportunities, risks, and impacts:
Today’s magical moment: my train from London to Edinburgh was stopped at
#york
due to an accident. In search of a café, I stepped out of the train station and saw the largest cherry blossom tree of my life standing literally just out of the station 🌸🌸
Last working friday of 2021 and i woke up to the wonderful news that I have two ethics/philosophy of AI papers accepted for publication at
#facct2021
@FAccTConference
Being a UK global talent & an asst prof doesn't make it easier. Dad's UK tourist visa application rejected cause the officer wasn't convinced that he'll go back home, despite all the proofs showing that his family, home, job are back in Tehran. F* disrespect & discrimination.
Thank you to our brilliant
#FAccT23
program chairs! Lilian Edwards (Newcastle), Jessica Hullman (Northwestern), Atoosa Kasirzadeh (Edinburgh) and Berk Ustun (UCSD). CFP out soon, submission deadline expected late Jan.
@Dr_Atoosa
@berkustun
@lilianedwards
@JessicaHullman
Our paper with
@IasonGabriel
is now available open access in the journal “philosophy & technology.” We provide a philosophical analysis of the value alignment problem for LLM-based conversational agents 👇🏽
The Open Access version of "In Conversation With AI: Aligning Language Models with Human Values", co-authored with
@Dr_Atoosa
, is now live in the journal Philosophy and Technology today:
Check out this🧵for details...
Excited to contribute to a new Agenda Paper on LLM Safety & value alignment. A critical question is what values, and whose values, should we encode in our LLMs? This paper calls for more research on both the technical & social aspects of this important issue. 🧵⬇️
I’m super excited to release our 100+ page collaborative agenda - led by
@usmananwar391
- on “Foundational Challenges In Assuring Alignment and Safety of LLMs” alongside 35+ co-authors from NLP, ML, and AI Safety communities!
Some highlights below...
This is such a disappointing news, coming from a company which gave many of us hope about the in-house impact of AI ethics-safety team in operationalizing Responsible AI principles.
@nytimes
@VivianHYee
Second, what exactly is "breaking" in this news? Do you understand what
#womanlifefreedom
uprising is about
@vivianhyee
? From this piece of writing, it's clear that you don't. Remove this piece of falsehood and next time, write about topics you have sufficient knowledge of.
Academic colleagues: The brutal behavior of armed guards in Tehran’s Sharif University, the Alma master of the first female Fields medal winner in the world Mirzakhani, is beyond words. Innocent students are being bitten and killed. Raise awareness. Be our voice.
Can you imagine the US government sending armed guerrilla forces to Boston to surround Harvard and MIT and start killing students? Well, that is exactly what is going on in Tehran, Iran as I tweet this. The world media seems to be deaf and blind!!
#دانشگاه_شریف
#مهسا_امینی
Dear academic and activist colleagues,
Here is a statement calling for (more) transnational solidarity with Iranians fighting for freedom and justice
#MahsaAmini
#مهسا_امینی .
Sign it if you like:
excited that our paper with
@IasonGabriel
on the value alignment of LLMs is making it through the heart of conversation on Responsible AI. The paper will be available online in Philosophy & Technology next week. I’ll write a 🧵 then sketching our argument and the next steps ahead
We are committed to responsible AI research with real-world impact. 🌎
Some of our recent work has focused on aligning a system’s values, understanding behaviours and reducing harm.
Want to find out more? Here are 3️⃣ papers to explore these ideas. 🧵
We are announcing new grants for research into systemic AI safety.
Initially backed by up to £8.5 million, this program will fund researchers to advance the science underpinning AI safety.
Read more:
The preprint of our
#facct2021
@FAccTConference
paper (co-authored with
@andrewthesmart
) "The Use and Misuse of Counterfactuals in Ethical Machine Learning” is now available.
ACM library:
Arxiv:
I had lots of fun chatting with
@JohnDanaher
about the value alignment problem for large language models. I share some thoughts about value alignment problem as an important framework for talking about ethical and engineering-safety questions 👇
New episode - 109 with
@Dr_Atoosa
: How Can We Align Large Language Models Like GPT with Human Values? We talk about different strategies for ensuring value alignment and the role that the philosophy of language can play in addressing this problem
Thanks to
@apa_blog
for inviting me to write a piece about the feminist uprising in Iran and to the fantastic artist
@RoshiRouzbehani
for allowing me to use one of her magical illustrations. 👇🏻
Really excited to be back in Montreal and attend
@AIESConf
as a panelist and presenter of two co-authored papers.
Paper (1) examines a safety concern in RL-based recommender systems.
Paper (2) offers a comprehensive review of the direct risks of generative text-to-image models.
🚀 Just in! The lineup for
@FAccTConference
is LIVE with a list of outstanding CRAFTs, Tutorials, and accepted papers! 🎉 Check it out: . See you all in Chicago in June!
#FAccT2023
I truly enjoyed being a panel discusstant on
@AlJazeera
alongside
@GaryMarcus
and
@rameshmedia
. We talked about "the letter" and regulation of generative AI models. Listen to our conversation here:
I recently spoke with
@samschech
from
@WSJ
about the pressing need to find and cultivate common grounds among different communities concerned about the risks of AI. Read the full report👇🏻
@andrewthesmart
and I will present our paper "The Use and Misuse of Counterfactuals in Ethical Machine Learning” today at
@FAccTConference
#facct2021
at UTC+0: 8:00PM. Looking forward to get your questions.
Open access for download here:
The preprint of our
#facct2021
@FAccTConference
paper (co-authored with
@andrewthesmart
) "The Use and Misuse of Counterfactuals in Ethical Machine Learning” is now available.
ACM library:
Arxiv:
2 women died last week. 1 in Scotland (where I live) at 96
#QueenElizabethII
. One in Tehran (where I was born) at 22
#MahsaAmini
#مهسا_امینی because morality police beat her for not having hijab the way police wished. These photos represent Iranians fighting for freedom and life.
Reading a good book about AI is great, but reading one from a colleague and a truly inspiring person you know? That's an experience like no other. Can't wait to dive into the physical copy of this book✨
Hey everybody, since this site is as stable as a shipwreck on stilts, I should probably mention now that we at
@CentreTMFutures
@UoE_EFI
are launching a new MSc in Data and AI Ethics
@EdinburghUni
and applications will open soon for Sept 2023 start! 1/3
I understand all the horrible things going on with Twitter and the harms the clown CEO has brought to many communities and epistemic practices. But, despite this, I want to flag that Twitter is one of The few successful platforms for the voices from Iran's uprising. 1/2
As a matter of empirical fact, we are living in a world where some say they advance AGI for benefiting all of humanity while even basic tech remains unused in distributing food to those facing famine and starvation. Who's gonna ensure that AI's advancements would benefit all of
In honor of International Women's Day on March 8, I’d like to celebrate the legacies of 4 of my favorite Iranian women, among many others. Let’s recognize their sociopolitical achievements and strive to uphold the high standards they set for what it means to be a human 1/5
Thanks to
@SOFIresearch
for inviting me to give an adhoc talk about Iranian women and
#MahsaAmini
during their academic conference event. So thankful to all the attendees for their solidarity 🙏🏻
We thank Atoosa for sharing her experiences as an Iranian woman with us! We are grateful for the information about the situation of women in Iran. And we would like to express our solidarity!
@AgnesCallard
I couldn't disagree more with this. This essay would be better titled 'A case against travel,' given that it presents a strikingly limited perspective. Travel makes many people into the best version of themselves while convincing them that they can continuously improve themselves
Giving my first 2021 talk
@FAccTConference
#facct21
(accepted abstract). Listening to my own pre-recorded talk and oh How much I miss physically being in a conference.
#explainableai
Preprint here:
I was thrilled to join the “think for tomorrow” panel at
@szigetofficial
. We had a packed room and discussed social, legal, and political implications of artificial intelligence. Thanks Budapest and the island of freedom for being an incredible host!
Such low point
@nytimes
has reached by publishing a blatantly racist opinion piece by Thomas Friedman titled “Understanding the Middle East Through the Animal Kingdom.”
If you're at
@FAccTConference
, join us on Wed 14:15 for an exploration into the governmental online targeting. We explore its implications & potential reconciliation with democratic principles. A case study on UK government included.
#FAccT2023
Paper:
How can conversation with AI systems be aligned with different values?
In this new paper with
@IasonGabriel
, we examine this question through the lens of pragmatics and philosophy.
How can conversational agents be aligned with human values?
New research from
@Dr_Atoosa
and
@IasonGabriel
explores this question using philosophy and linguistics:
Very excited to share the final line-up and program of our upcoming conference on the Philosophy of Deep Learning!
Co-organized with
@davidchalmers42
&
@De_dicto
and co-sponsored by
@columbiacss
's PSSN &
@nyuconscious
.
Info, registration & full program:
My name is Otto Circuses Order!! With this fancy auto-transcription of my name, I feel like
#ItsTime
to become a supremeleader of a sort (in
#starwars
or elsewhere) and bring some order to the circuses.
#singularity
is coming 🤖
🗣non-western AI ethics refers to the constellation of problems, concerns, and methodologies that arise from the use of AI in non-Western societies. I hope it’s clear that NW-E includes, but goes beyond, applying non-Western ethical theories to AI.
📢 Searching for academic/non-academic researchers who work on non-Western AI ethics. Reason? Inclusion in a couple of forthcoming AI ethics events to increase the diversity of perspectives. Please refer me to names either in the comment or in private message. Thank you 🙏
Just a quick note to thank all my students and colleagues who send me messages of empathy, sympathy, and solidarity ❤️🩹. Your messages and our conversations help living through these days as an Iranian abroad.
#WomanLifeFreedom
#MahsaAmini
Amazing talk by
@vardi
at
@UofTEthics
#uoft
on the
#societal
challenges of
#AI
and the importance of public policy in addressing the challenges. I hope more and more computer scientists get engaged in the debate.
It has been over 4 months since the onset of the Iran protests and unrest
#WomanLifeFreedom
. The demonstrations have developed in an organic and decentralized manner. I’ll provide a more comprehensive opinionated view shortly, but to summarize: the situation is evolving.
Happening this Sunday: The leading candidate in the Social Change category at
@RecordingAcad
(
#GRAMMYs
) is “Baraye,” a protest anthem by Iranian composer
#ShervinHajipour
inspired by the mass protests in Iran after the killing of
#MahsaAmini
. Listen to it 1 more time. #مهسا_امینی
Curious about how different
#ethicsofAI
projects relate to each other? Then, read our paper (joint work with the excellent
@cvklein
) forthcoming at the proceedings of
@AIESConf
Preprint here:
I've been blessed to learn about Israel/Palestine through the lived experiences of my Israeli and Palestinian friends - Jewish, Christian Arab, Muslim Arab - over years. Their shared commitments to human rights has shown how possibly we can co-exist in peace and goodness. 1/2
اولین باری که نشستم به بخشی از عمر حرفه ای نگاه فارسی طور کردم در یه گفتگوی بی شیله پیله با دوست خیلی عزیزم سروش . هیچ ویراستاری روی مکالمه انجام ندادم بعد ضبط که جنس اصل بیاد بیرون 😅 چون ترجمه فلسفه علم فارسی هیچوقت نخوندم کلمه فارسی کم میومد که درست میشم کم کم 🙃
Podcast | Episode 6
Soroush Marouzi chats with Atoosa Kasirzadeh about the roles of mathematics in empirical sciences.
Spotify:
Google Podcast:
Podbean:
@AmandaAskell
To benefit all of humanity, let’s just agree that AI ethics and AI safety ask many joint questions And that the two communities should join forces as soon as possible. The communities for various sociological reasons got separated imho but peace time is needed
Today marks the 40th day from
#MahsaAmini
’s death. This is a powerful video showing thousands of people walking from
#Saqez
towards Mahsa’s cemetery. There has been many restrictions by authorities in place to prevent any gathering over Mahsa’s grave on this day.
#مهسا_امینی
@AIESConf
In "Typology of Risks of Generative Text-to-Image Models", we investigate the direct risks and harms associated with modern text-to-image generative models through a comprehensive literature review.
preprint --->
Are you a qualitative researcher interested in data/AI governance and social trust? We’re hiring a postdoc for our
@tas_hub
project on responsibility for autonomous systems; apply by 17 Jun to work w/
@nayhasethi
@EdinUniUsher
& our interdisciplinary team!
🗓️Technomoral Conversations event alert🤩
Join us on the 29th of March at 5pm for a conversation on the future of Responsible AI between two leading voices in Europe and Scotland,
@vdignum
and Albert King, Chief Data Officer for Scotland. Book here 👇
Calling all experienced reviewers! This year we're opening up self-nominations for reviewing
#FAccT2023
. Sign up now and help shape the future of the field!
From Australia to US and back again in a few hours. Thanks to
@jhimmelreich
and
@JCCHerington
for organising this workshop.
Our paper (wiith
@cvklein
) is forthcoming at the proceedings of AIES 2021, and you can read its preprint here:
A cool framework for thinking about problems in AI Ethics:
@Dr_Atoosa
presents "The Ethical Gravity Thesis" (with
@cvklein
). Video from 2021 Upstate Workshop on AI and Human Values
Here is a video of a talk in English recorded on Oct 1st 2022 about what’s happening in Iran. This talk provides a short historical context for understanding the significance of the protests in Iran. 👇🏻
#مهسا_امینی #زن_زندگی_ازادی
🗓️Join us for an in-person 🤩 lecture from
@Abebab
that will be free and open to all (within room capacity), on "The Limits of Fairness" in AI ethics. Co-sponsored with
@aies_edinburgh
on 13th June at 5pm in Edinburgh
@UoE_EFI
Details and registration: 👇
5. Fucking go for it. Whatever it is. Just try. Maybe you make it. Maybe you don’t. The result isn’t the point.
Your journey, your growth, is the important bit. Don’t let anyone convince you shouldn’t try. Their fear is theirs, it’s not yours to carry.
Iran’s Food & Drug Administration has ordered pharmacies to force their female staff to wear black veils at workplace. Iranian men are mocking this order and supporting their female colleagues by wearing hijab.
Compulsory hijab is the main pillar of a religious relationship.
Persian Gulf International Airport in Iran reported a heat index of 152°F (66.7°C) today at 12:30 pm.
Those are intolerable conditions for human/animal life.
@GiorgiaMeloni
Iran🇮🇷 & Italy🇮🇹 have the same tricolors on their national flags. Iranian women are losing their lives fighting for freedom. This flag 👇🏻 has become the symbol of this fight with 3 words of life, freedom, and woman at its centre. Hope you preserve freedom for everyone in Italy.
We are excited to invite you to the first event in the new series: Technomoral Conversations
@UoE_EFI
Join us,
@ShannonVallor
@Dr_Atoosa
and
@IasonGabriel
on the 7th of Dec for a discussion about the ethics of large language models. Book here:
An extremely rich CRAFT session at
@FAccTConference
on the multi-dimensional impact of generative AI on artists. The message: "Don't take our works without our consent and earn money from it."
At a CRAFT session *with actual artists* talking about AI Art.
A point being made now is that art has always been shared, but that's different from it being cannibalized by corporations without artist consent.
#FAccT2023
Programming is a powerful way of thinking. Your job is to enhance the productivity of thinking by a safe automation of its sub-components. It is not your job to stop a powerful way of thinking on the premise that a trillion-dollar corporation's network of computational
Jensen Huang, CEO of Nvidia, argues that we should stop saying kids should learn to code.
He argues the rise of AI means we can replace programming languages with human language prompts thus enabling everyone to be a programmer.
AI will kill coding.
This year, I'll be a lecturer at
@LMU_MCMP
summer school on mathematical philosophy for female students. The summer school is entirely online. Details for application below 👇
Students have a few more weeks to register for this year's MCMP Summer School on Mathematical Philosophy! Great line up of talks, and only €20 (+waivers available) find out more here:
Had the pleasure of leading 3 workshops in spring23 with the goal of surfacing new approaches to human flourishing with AI, bringing together expertise, insights & provocations from groups often left out of high-level discussions of AI & our futures. full report & exec summary👇🏻
AI-based technologies are often designed for a specific group of people, rather than with and by that group. How can we shift from a design-for culture to a design-with-and-by culture?
Read about this & more in our New Perspectives on AI Futures Report ▶
Happy international women's day
#InternationalWomensDay
#IWD2021
to all the hopeful, brave, and strong women who fought and are fighting against descrimination in the world. I grow up in light and memory of your inspiring past/present stories everyday.