Looks like today is a day to share research and good news!
I am a big fan of
@BigDataSoc
and I am so happy that it will now be the home of our work "Diversity in Sociotechnical Machine Learning Systems", joint work with my amazing collaborator
@sinafazelpour
, link & summary 👇
📢 Our paper, "Diversity in Sociotechnical Machine Learning Systems", (w
@mariadearteaga
) is now out in
@BigDataSoc
, and it's open access:
See this thread, where I discussed the main points of the paper (wrt. an older version) 👇
I'm a Doctor! 👩🎓🎉
I never imagined I'd be defending my thesis on Zoom during a global pandemic, but it still feels quite amazing. I'm so thankful for all the people that came into my life during these six years, and for the people who supported me along the way.
In today's example of how *not* to do data visualization, Colombia's president has chosen a bar graph to compare aggregated murders during 8 years of the past gov vs. during 2 years of his gov.
Nos duelen los muertos que deja la violencia producto de narcotráfico y terrorismo. Entre 2010 y 2018, nuestro país vivió 189 homicidios colectivos, y entre 2019 y 2020, 34 hechos de esa naturaleza. Seguiremos combatiendo a disidencias FARC, ELN, Clan del Golfo, carteles y otros.
I am on the academic job market!
My work is on human-centered machine learning (ML), and focuses on issues of fairness, accountability and sustainability when using ML for decision support.
Please reach out if you think we may be a good fit!
A few months ago, a Colombian police officer raped a woman, and when protesters vandalized the police station where she was raped, people decried the graffiti more than the rape.
That's when I understood the human cost of being outraged about damaged property.
This has been a key feature of my academic life. I have a joint PhD in Machine Learning & Public Policy; I had to be admitted to *both* PhDs and I had to complete the requirements for *both* PhDs. And yet this has so frequently led to people dismissing me as "less technical".
I've been explaining how, for Computer Scientist women, picking up more skills means the others are erased & you're further demeaned. I started programming at ~10. Trained my first model when I was ~21 (without ML packages). Worked in "AI" for ~15 years. I have a doctorate in CS.
Very happy to be a recipient of the
@GoogleAI
Award for Inclusion Research! And really looking forward to working on my proposed research--how machine learning can be helpful for amplifying advocates in algorithmic decision support.
We’re happy to announce the recipients of the 2020 Award for Inclusion Research, a global program that supports academic research in computing and technology that addresses the needs of underrepresented populations.
I am thrilled to announce that I will be joining
@UTAustin
this summer as an Assistant Professor in the Information, Risk and Operation Management Department at
@UTexasMcCombs
. So happy to have found such a fantastic interdisciplinary home!
There are many things I love about being faculty at a university with >20% Latinx students, but my favorite is definitely being called Profe on a regular basis 😍
The dissociation I'm feeling on Twitter is a painful reminder of the US-centrism of academic communities focused on 'justice' & 'ethics'. 100s killed in Palestine, 100s forcibly disappeared in Colombia. And the conversation here continues as if nothing was happening.
We are very excited to announce the 3rd NeurIPS workshop on Machine Learning for the Developing World (ML4D)!
This year’s theme is ‘Challenges and Risks’: What can go wrong with ML4D technologies? How do we tackle this?
As our Iranians colleagues tell us about the lack of support they've felt from the rest of us, it may be a good time to reflect on how the lack of support extends beyond this. The pervasive barriers they face bc they're from an embargoed country are invisibilized in academia🧵
I am emotionally exhausted. For months I didn't know how the work visa bans would affect me, and now my SO’s visa is at stake.
But it's disturbing to see folks implying that the new bans cross an ethical line. As if being “highly skilled” makes targeting us worse.
After a successful mid-tenure review (yay!), I am eligible for junior faculty leave, and I am considering visiting researcher positions. I’d love to spend a few months collaborating on human-AI collab and/or alg. fairness. If we may be a good fit, please reach out!
Prospective PhD students: Are you interested in interdisciplinary research on machine learning, human-AI complementarity, ethics, network science, computational social science? Consider applying to be a part of the Computational Data Science group at
@UTexasMcCombs
.
The use of "ethical approach" in this description is a perfect example of why AI ethics cannot and should not be reduced to a problem of AI bias, and then further reduced to a problem of disparate performance across groups. There is no "debiasing" an inherently transphobic task.
As a grad student, whenever faculty said that working with students was the best part of the job I mostly thought: *we* are the best part of your job? 🤔
As faculty, working with students is indeed amazing 🥰
So OpenAI's plan to "ensure that artificial general intelligence benefits all of humanity" is to build the exact same technology everyone else is building but just be very optimistic about it? Because that'd explain a lot
Techno-optimism is the only good solution to our current problems.
Unfortunately, somehow expressing optimism about the future has become a radical act.
It's hard to put into words how out of place I felt the first year of ML PhD courses. I once asked a TA for supplemental material to cover gaps and was told in front of the entire class that doing a PhD meant we should find material ourselves.
Help your students know they belong.
@akbarpour_
When I taught first year courses, it was very clear to me that a lot of the showmanship exhibited by first year grad students came from those who took the classes before, and this made other students feel inadequate. Important to warn students so they aren't discouraged!
The fact that during BLM protests one of the most powerful voices in ML + many others have chosen to gaslight, mansplain, and tone police early career Black women is so telling of what needs change in the ML community.
Folks, this is a time for listening and learning.
It has not been surprising but informative to see the types of things
@ylecun
's post engenders, and what his followers say. Like flipping the whole reality, as if we're not the ones with career consequences. Next they'll probably say we're the ones who are treated great by police
Our
#CHI2020
paper “A Case for Humans-in-the-Loop: Decisions in the Presence of Erroneous Algorithmic Scores” is now on arXiv!
Joint work with Riccardo Fogliato and Alexandra Chouldechova. Short thread 👇
📢I am recruiting a postdoc! If you are interested in human-ML collaboration in healthcare, overcoming limitations of observed labels + mitigating risks of self-fulfilling prophecies, and working closely with stellar domain experts, consider applying!
Having started my research career in Colombia, I am aware of how inaccesible conferences are for many, and what the impact of this is. It is time for major conferences to start moving around the world (and not just the wealthy countries of the world).
@cfiesler
While not serving alcohol seems perfectly reasonable, my problem with "lets get rid of alcohol" as a solution is that we should really be able to have a drink with colleagues without risking being sexually harassed/abused. I'd prefer we get rid of the sexual predators.
Join, Keynote speaker Maria De-Arteaga as she talks about "Three Frontiers of Responsible Machine Learning" for LatinX in AI at
#ICML2022
this July in Baltimore, Maryland.
For details, please visit
#LXAI
#ICML2022
The assumption of a “utility-fairness tradeoff” is still way too common in conversations around algorithmic fairness. We should instead be talking about a utility-fairness relationship. This is a summary of key aspects of this relationship🧵
1/
Today I am both so happy to have received my first shot, and so frustrated to know that my parents are not even eligible yet. The global disparities in access to vaccination are appalling.
This week many of us are thinking (and tweeting) about the wonderful things
#NeurIPS19
left us. Experiences, connections, new knowledge...those are all things that are being denied to folks who couldn't obtain a Visa. Lets be thankful, but lets not forget that change is needed.
This year's
@FAccTConference
offers:
1⃣Discounted registration for countries listed by ACM as economically developing (based on World Bank GNI).
2⃣ Internet bandwidth grants to cover extra costs of internet access.
3⃣Carer grants to cover extra costs of caretakers where feasible.
Registration is now **OPEN** for ACM FAccT 2021! The conference will be held March 3-10.
Early bird pricing is available through February 18. 🤩
Please check our website for details on volunteer positions, Internet Bandwidth & Carer grants and more!
After 2 years of work, our extended analysis of our CHI’20 paper is ready and the preprint is available! And we still want to make A Case for Humans-in-the-Loop (joint w/ Riccardo Fogliato &
@achould
)🧵
A few days ago, 🇨🇴 ex-president Uribe called for the police to shoot at protestors. That is exactly what is happening now. Meanwhile,
@NYUWashingtonDC
is hosting him tomorrow for an event titled "Sustainable Colombia"(!). Shame on you,
@nyuniversity
.
I don't know who needs to hear this, but "it's a pipeline problem" doesn't mean that the problem is just upstream. It also refers to the portions of the pipeline that you gatekeep.
As early-career researchers, we are told to minimize our service load. There is value to that, but for members of underrepresented groups service is often about creating spaces where we can exist. Not rewarding service has a disproportionate impact.
I'm going to tag along and share news :) I am also now a part of the
@FAccTConference
Executive Committee! Being DEI chair was very rewarding, and I am incredibly happy to continue to serve a community that I care about deeply, along with folks that I admire so much.
Have I mentioned yet that I'm now on the executive committee for
@FAccTConference
? Very happy to keep contributing to this vitally important community with such a dedicated team. Feel free to reach out if you have thoughts for how FAccT can keep getting better.
It's a relief to read this and I'm so thankful to those who fought so hard to protect others at such a high cost. They shouldn't have had to do it.
But let's not forget that the system failed them, we need to structurally change it, and Walter is not the only one who needs to go.
As many others, I just got COVID for the first time. Tweeting this to contribute to the awareness that we are still in a pandemic and expectations of "back to normal" are unreasonable. Also, I want to echo the fact that calling this "mild" is a mischaracterization.
If anything, this is a moment to learn about empathy and solidarity. You are angry about this? Then be angry about *all* of it. Stop being complicit with the xenophobic system that got us here in the first place.
It has been very inspiring to see the impact that
@timnitGebru
,
@jovialjoy
and
@rajiinio
have had, both through their research and their tireless efforts to get others to listen. This is such a great example of how CS research can inform policy.
I met with
@hannawallach
at NeurIPS'17 and it was very valuable. This year, I started meeting with students who are more junior than me.
No matter how junior you feel, you are always senior with respect to someone--reach out to senior folks and pay it forward✨
This was so worthwhile. I met with ~30 students and postdocs (or equivalent career stage) over the week, plus a handful of assistant professors (or equivalent) too. 1/3
What is happening to Timnit at Google is *very concerning* and the
@FAccTConference
community ought to pay attention. Timnit is a trailblazer, and she's now bringing us to a much needed conversation about the role of tech companies in shaping and constraining AI ethics research.
Annual reminder to academic friends: never tell a graduate student of color, especially a woman, that getting a job should be easy for them because of diversity initiatives. When you do that, you erase their hard work, you slight them, and you ignore their struggle. Don’t do it.
This first week of class--going back to in-person teaching while Austin is Stage 5 and we cannot require masks nor vaccinations--has been beyond difficult. 1/
Basically everyone I follow on Instagram to stay informed of what is happening in Colombia (friends, activists, + non-mainstream media) is reporting that their content and stories are being deleted. Is this the gov paying bots to report? Rogue algorithms? A tailored policy by FB?
Next week (Nov 19-20) is the 2020 Global Analytics Summit: Ethics in AI! Thanks to the support of
@UTexasMcCombs
and
@UTGoodSystems
registration is free. We have a tremendous line-up of speakers, checkout the program and register here:
We received a record number of 120 submissions to ML4D NeurIPS 2019 Workshop!
2017 (inaugural): 26 submissions
2018: 64 submissions
2019: 120 submissions
The ML4D community is growing! 👏
📣New preprint! “Leveraging expert consistency to improve algorithmic decision support”, with
@achould
and A. Dubrawski. We propose the use of influence functions to learn from historical expert decisions and mitigate limitations of observed labels.
Excited about these news + the launch of the Machine Learning Laboratory, a multidisciplinary home that brings together ML research happening across campus:
📣New preprint! “Leveraging expert consistency to improve algorithmic decision support”, with
@achould
and A. Dubrawski. We propose the use of influence functions to learn from historical expert decisions and mitigate limitations of observed labels.
📢 If your are at
#NeurIPS19
remember that tomorrow is ML4D! Stop by to hear from our amazing invited speakers
@Bomu
,
@kentarotoyama
,
@Abebab
,
@rajiinio
+ poster session + contributed talks + panel discussion. Room West 121+122. Thread 👇
Taking a break from harvesting snow for my waterless, heatless home to say that this is an very concerning development of an already concerning situation. Google has hired *so many* AI ethics researchers, to then do...this.
I had a great time kicking-off my
#facct2021
attendance this year with this tutorial! Stay tuned for accompanying resources. I'm feeling so lucky I get to collaborate with
@sinafazelpour
🤩.
In Stream Two -> "Sociocultural Diversity in Machine Learning: Lessons from Philosophy, Psychology and Organizational" by Sina Fazelpour and Maria De-Arteaga
#facct21
What is most heartbreaking and disgusting about this is that we are finding out via Twitter because
@UMichCSE
and the academic community have failed the victims, and this man was (is?) about to obtain tenure = a lifetime of access to students from a position of power.
Walter Lasecki assaulted me. He pinned me against a bar, he put his hand up my skirt and his fingers in my underwear. He grabbed and twisted. He leaned in and insisted that I sleep with him. “It could be innocent,” he said.
Today Colombia is on strike. Among other things, we are telling the government that we are not going back to the war that they are trying to bring back. We demand a government that respects human rights.
#21N
(T-shirt features peace agreement handshake)
I had so much fun in this joint interview w
@minlee_hci
. Among other things, we talked about issues with ML's "move fast and break things" approach, the way our work relates to each other, and our paths to where we are. So happy to have her as a colleague!
📢Tenure-track job postings 📢 My department is hiring!
First position: Use-inspired AI. We seek candidates pursuing a use-inspired AI research agenda in the context of important business, organizational, and societal challenges
I just gave a talk at the launch of the data science undergrad at
@UniJaveriana
and I'm so impressed! The program is joint between the Science, Engineering, Communications, and Econ Schools, and the whole event was grounded on societal responsibility and critical perspectives.
This week
@UTexasMcCombs
and
@UTGoodSystems
are hosting a conference on Explainable AI, bringing together an amazing lineup of speakers and panelists. Registration is free and open to the everyone, check it out!
As
#AI
and machine learning techniques evolve, their value to businesses are increasing. Learn more about the implications for organizations at the 2021 Global Analytics Conference on Explainable AI, hosted online by
@UTexasMcCombs
. Learn more:
In the coming weeks I'll be giving (virtual) talks in Panamá, Australia, and Colombia:
Jul 17:
#FundaProMat
(en español!) 🇵🇦
Jul 22/23 (depends on your time zone):
@hmi_anu
Data, AI and Society Seminar 🇦🇺
Jul 23:
@WiDS_Conference
Bogota, Keynote (en español!) 🇨🇴
Details soon!
I talked with
@nytimes
for this piece on the new NIST report (thanks to
@CadeMetz
for reaching out!). On this thread I elaborate a little on my perspective 👇
In our 'Bias in Bios' work (FAccT'19) we studied risks of bias in occupation classification. This right here is precisely the type of gender bias we warned could arise.
Folks, I want your recs! 😊 I am wrapping up my intro to programming course (Python), and I want to provide students with resources to continue learning and progressing. What are your favorite coding resources? 💻
(Competitions, projects, interview preps... anything counts!)
This "sexual harassment classifier" is a great example of techno-solutionism crap.
You know what's a great way to detect sexual harrasment? Listening to victims!
Victims are not reporting in your org? The problem you need to solve is not "automated harassment detection".
Microsoft Purview is a new product aimed at enabling companies to fulfill their corporate compliance by analyzing emails among other things. The list of issues it can report based on analyzing employee emails is completely dystopian.
False positives here will likely ruin lives.
During my PhD I was lucky enough to do an internship with Adam, and as faculty I've had the pleasure of co-advising interns with him, and I can't recommend this enough! Adam is an amazing and caring mentor, and the MSR New England Lab is a research wonderland. Apply!
If you are a PhD student looking who would like to do a summer internship with me on Program Synthesis (teaching computers to code) or with any of our other fantastic ML folks at Microsoft Research New England, please apply here:
Wow, Twitter ❤️ you really are amazing at showing love, thank you! The hat has been very popular :) it was a surprise by
@BenBoecking
and I plan to just be here social distancing, drinking prosecco and wearing a tam (may wear it to Zoom meetings, too).
I just boarded a
@Delta
plane and the hostility I experienced for not wanting to opt into facial scanning, and for telling other passengers that it is not mandatory, was really something. So much for consent in data collection.
I don't think there's anything wrong with making policy based on the intersection of what's measurably good for the economy & ethical guidelines (although I think that migrating is a human right). But saying this is worse on moral grounds is dehumanizing to other immigrants.
I don't recall any DEI initiative that specifically considers the experience of those from embargoed countries (as someone who has been DEI chair and served on hiring committees, I obviously share responsibility).
My department is hiring! We are hiring in “use-inspired AI”, and seek candidates pursuing a use-inspired AI research agenda in the context of important business, organizational, and societal challenges. Read more here:
This story is terrifying as is, and it's extra-terrifying that if she didn't own an iPhone she wouldn't have realized what was happening. It's v problematic that the "safety feature" that protects(?) people from being stalked with this product is only for iPhone owners.
There are countless fellowships, internships and other opportunities that they are excluded from, and their ability to travel internationally is often severely limited. And yet, nobody seems to be aware of this or have an issue with it.
Undergrad student &
@MSFTResearch
intern, Myra Cheng, will give a spotlight on our work "Social Norms Bias: Residual Harms of Fairness-Aware Algorithms" (joint w/
@adamfungi
&
@LesterMackey
). Come watch her talk!
You can also visit her poster at "Socially Responsible ML" on Sat!
This is a symptom of a deeper problem. I'm not calling for the same communities to start talking about different things, but for those communities to change and become truly global.
Very happy to share our
#IJCAI2021
paper, led by
@UTexasMcCombs
PhD student
@ruijianggao
. We propose methodology for human-AI collaboration under bandit feedback, and considering heterogeneous expertise across humans. In this 🧵he provides a fantastic summary.
However, it is encouraging to see how badly they underestimated how resolute the AI ethics team that
@mmitchell_ai
and
@timnitGebru
built is. And I cannot wait to see all the amazing work that these stellar researchers will do wherever they choose to go next.
In addition to Visa barriers--which disproportionately affect researchers from developing regions--the financial burden is also unequally distributed.
If we want to be a diverse and global community, our choice of conference locations needs to consider the whole community.
I realize that all countries have their shortcomings, but the issue with having the same country host 3 years in a row is that the exact same group of people are discriminated against each time.
Very excited to be co-organizing this event with my soon-to-be colleagues at
@UTAustin
!
With keynotes by
@ecekamar
&
@manojsaxena
+ many fantastic speakers, it'll be a full day to learn more about AI ethics📝
Event website:
At some point we should talk about that new-ish journal publishing lots of work on responsible computing & charging $8,900 publication fee. I see many scholars I respect recently publishing there, and I truly don't understand why we'd want to support such a predatory venue.
Colombian police says they'll use facial recognition to identify those who cover their face at protests. What alg is it? How will the output be used? Is it valid evidence when accusing someone of a crime? There's a growing need for algorithmic oversight and accountability in 🇨🇴.
#VIDEO
El helicóptero Halcón de la Policía de Bogotá estará funcionando con un dispositivo que permitirá identificar los rostros de personas que participen en actos de vandalismo durante el paro nacional
#Ma
ñanasBLU.
The FAccT paper deadline is behind us, but there's plenty ahead, including the inaugural
@FAccTConference
Diversity and Inclusion (DEI) Scholarship!
Does your research seek to understand the impact of and advance DEI initiatives? Consider applying to become a FAccT DEI Scholar!
📣 New
#facct2022
paper “Justice in Misinformation Detection Systems: An Analysis of Algorithms, Stakeholders, and Potential Harms”.
Led by 1st year
@UTexasMcCombs
PhD student
@TerrenceNeumann
(say hi to him in Seoul!), and joint with
@sinafazelpur
🧵
This week we would’ve been in Hawaii presenting our work at
#CHI2020
. Now, we are finding other ways to share our findings. Here is a blog post we wrote (w/ Riccardo Fogliato &
@achould
):
Dear US friends, there's hope for us banana republics, don't give up 🎉
In all seriousness, I hope this brings us one step closer to the construction of historical memory that Uribe has denied the victims of paramilitary and state violence.
Álvaro Uribe was Colombia's political giant, the untouchable 'Teflon president.' Now, he faces jail or house arrest amid a fraud and bribery investigation, a first for a modern Colombian president. This looks like a landmark moment.
As someone who currently sits in an IS group, I think there's lots of interest+confusion on how to welcome ML folks in Bschools and ISR. This thread is the story of how ISR lost the opportunity to claim a top ML researcher doing exactly the type of work the field strives for.
Ostensibly, this should be a golden era for "information systems research". Between AI & fintech, new info-tech is impacting business and organizations everywhere. And yet the field that claims this turf is nowhere to be found. What's going on? (1/n)
Question: When graduating from a PhD in the US without greencard/citizenship, how important in practice is the STEM OPT? Do people often go into the extension period? In what circumstances? Please share your knowledge & experiences! 🙏
#phdchat
@AcademicChatter
👩💼👩🔬No se pierdan nuesta segunda charla mañana martes 24 de Noviembre a las 6pm ET. Riesgos de discriminacion en inteligencia artificial, por la profesora Maria De-Arteaga de la Universidad de Texas - Austin
@mariadearteaga
@UT_IMProgram
@ut_orie
En vivo:
Our
#CHI2020
virtual talk "A Case for Humans-in-the-Loop: Decisions in the Presence of Erroneous Algorithmic Scores" is now on the
@sig_chi
YouTube channel (and so are all the other CHI talks!)