Welcome (back) to Parity!
This article is reposted from the Parity Substack. Follow us there for updates.
Hello, world!
Welcome back to Parity, now coming from the brand new team. We took over for Dr. Rumman Chowdhury while she’s off solving some of the world’s most pressing (and challenging) algorithmic issues at Twitter. Rumman’s still with us as our lead investor and board member, helping us immensely as we grow. This is the first newsletter we’re sending from the new team, and we plan to make it a place where you can keep track of all the new and interesting things we’re up to at Parity (2.0).
Starting a company during a pandemic is no easy feat. But for Parity, this moment has been the gleeful culmination of years of scholarship, hard work, perseverance, and a lot of good luck. We’ve just returned from a week of covid-safe socializing together in Phoenix, Arizona, the site of the very best Airbnb mega-home capable of housing the whole extended team we could find. Along the way, we learned a lot about each other, our goals, our values, and the product we’re building. We’d love to share some of the decisions we made together, the promises we made to each other, and invite you to follow along as we grow into the gold standard for responsible AI model governance available on the market today!
Meet the team!
Parity was first founded by Dr. Rumman Chowdhury, who was quickly acqui-hired into Twitter within its first year of operation. Rumman’s patent-pending NLP risk-identification technology is a foundational feature of the platform, and the inspiration for the next phase of Parity products we’re excited to tell you all about soon! But we’re not here to talk about our products yet. This post is about our team.
When Liz O’Sullivan, previously a cofounder and early employee at Arthur and STOP respectively, reincorporated Parity this past May, she set out to recruit the world’s best experts in algorithmic fairness. In October, Jiahao Chen, previously at MIT, Capital One, and most recently AI Research Director at JPMorgan Chase, joined as CTO. The team quickly grew to four full time humans, when Mike McKenna, who most recently worked on vaccine equity within Operation Warp Speed, and Aleksander Eskilson, coming from a career in the delivery of prediction models in healthcare, signed on alongside a bevy of advisers, investors, and well-wishers from academia and industry alike.
Our interests are far ranging, including activism for ethics and inclusion in AI, such as Mike’s work as an organizer at Queer in AI, and Alek’s work on the advisory committee at the Center for Practical Bioethics. We’re thrilled to also be joined by rockstar advisors like Bärí A. Williams of Facebook, and David Wieseneck of LetGo and Ollie’s Pets.
And of course, there’s Tank! Tank is our Chief Puppy Officer and Jiahao’s best friend. Here’s Tank enjoying some Arizona scenery. It was pretty hot…
An in-person offsite? Now?!
We’re building a remote-first company, and hiring nationwide to attract the best and most diverse talent available. A few times a year, we’ll get together to renew our personal relationships, and incorporate our truest selves into this blossoming company culture. In order to do it safely and inclusively, we surveyed the invited attendees for their post-vaccination covid safety preferences, and instituted the rules of the most-risk-averse among us. For this team, it meant that even though many of our staff feel personally at low-risk for themselves, they were obliged to obtain a clean PCR test prior to boarding flights. It also meant that some of our most coveted team-building activities (e.g. a drag show!) were off-limits. And for the safety of our bubble, all were asked to wear masks while outside of the pod we’d built at The Happy Roadrunner, our home for the week.
As we grow, our intent is to create a warm and welcoming environment for everyone, regardless of their biological, cultural, and ideological differences. We’re dedicated to the inclusion of marginalized communities among our team, which includes a community of folks who might otherwise not feel safe speaking up about, for instance, their disabilities. For example, our first offsite, as eager as we were to spend the time together, was not mandatory for anyone. Nor is any amount of in-person work. At Parity, we will never issue an in-office mandate, just one of the many promises we made to each other at the event.
Promises?
At Parity, we know that people are people first, and workers somewhere further down on the chain of important things. We’re small enough now not to have to worry about OKRs or KPIs yet, so instead we talked about the promises we’ll make to each other, and how they’ll eventually add up to shared trust. It’s been clear from the very first moment working with this team that we’re all the type to work harder and faster than anyone should, especially on a mission this important. So we wanted first to lay down some ground rules on how and when we’ll work together.
A dedication to inclusion starts with the understanding that we are different. For it to work, we must craft a culture that’s welcoming to (not just tolerant of) diverse perspectives. Not everyone is a 9am every day morning person, and that’s ok. Soon, we’ll have teammates with families, and we know that for them and many other groups, the unexpected is guaranteed. For this and other reasons, we committed to what we’re calling “Work Hours Anarchy”.
Recognizing that this may not be viable forever, or for teams with varied mandates, we leave it up to each teammate to put in the time whenever that time works best for them. Want to skip a Wednesday to drive your kid to a soccer tournament? Go right ahead! As long as we keep the important promises of ambitious but communally-decided deadlines and production schedules, we don’t intend to require any wasted face time for “butts-in-seats”. We update our Slack each working day with progress on our shared goals, but don’t feel the need for an official morning standup meeting, least of all with cameras turned on.
But, since some may find this degree of freedom unnerving (as they often do with unlimited vacation), we felt the need to call out certain parameters for work time. We have “no meeting Fridays”. We set the expectation that night-owl messages need not be answered immediately for the early-morning crew. We recommend and plan to enforce a 4 week minimum vacation. Feeling burnt out? Take more time! We have tight turnarounds and lofty goals, (and there’s still a pandemic going on…) so we’re building in mental health awareness from the very first day. And importantly, for people with uteruses, menstrual leave is oftentimes necessary, and will never carry a stigma for our team.
What we came up with
By the end of the week, the whole team was buzzing with excitement over our research agenda, product roadmap, and production schedule. We’ll do another post in the weeks ahead to announce the incredible platform we’ve been building, but for the remainder of this, our inaugural newsletter update, we’ll instead share the mission, vision, and values we promised to uphold for one another in defining who we are.
Our Vision:
End Algorithmic Inequity.
A good vision statement is one that no reasonable person could expect to accomplish in their lifetime, and that’s certainly true of ours. But we all come to Parity with the shared vision that this problem is worth working on, and will spend our time pursuing this goal.
Our Mission:
Blaze a trail that challenges conventions in the AI industry through collaboration, care, and rigor.
Compared to a vision statement, your mission should be actionable and clear. With ours, we wanted to convey that the degree of care and rigor applied in the status quo of AI is woefully inadequate, and that we intend to use our expertise to help shift this needle. We intend to demonstrate to the industry that Responsible AI is not only economically feasible, but something that will help enterprises save money and prevent reputational damage before it becomes a crisis.
Our Values:
We grouped these into three categories, stated as our shouldered responsibility to each of multiple groups.
Our responsibility to each other:
We are all people first.
By necessity, this requires a commitment not just to the D of DEI, but instead to agree that Diversity doesn’t start and end with the pipeline. Instead, we live by a framework of “Diversity, Equity, Inclusion, Support, and Development” (DEISD). Once you join the Parity team, our work becomes one of fostering creativity among diverse perspectives, while ensuring all people with different backgrounds have access to the unique support they need. This means affirming each other for our differences, and for all the good work we’ve done.
We owe each other more than just respect.
Too often, corporate cultures expect people to bring just the bare minimum of human decency to work. But the most exciting thing about Parity so far is the incredibly kind, generous, and humble team we’ve brought together. Because we are investing in DEISD as a core value, we are guaranteed to disagree. We promised to acknowledge that disagreements are not inherently combative or abusive, as long as they are approached with compassion and mutual admiration.
Our responsibility to our clients:
Our A+ team = A+ work.
We have a high talent bar, and it’s not something we plan to compromise. We believe in engineering excellence, and strive for the utmost quality in all we do. But importantly, we believe that knowledge is something one arrives at, following the hard work of collaboration, experimentation, and the incorporation of lived experience from a diverse group of stakeholders. By following these guidelines, we’ll avoid churning out software that claims to do more than it can, and intend to prove our claims substantively (and reproducibly) before release.
Trust is something earned, not given.
We ask our clients to trust us with the truth about their algorithms, and in doing so we must guarantee them a high degree of confidentiality.
We are ready to build this trust with our clients, whatever their needs are. Parts of our platform may function under zero trust scenarios where clients are only willing to share the most basic of information. Under scenarios with more trust, we offer our clients some legal mechanisms to maintain confidentiality and open up key conversations which might otherwise be considered too risky. As trailblazers, we will be asking questions and developing solutions which may be new to our clients, and we are committed to building a protected environment where these processes can take place with minimal security risks.
Our responsibility to the world:
We won’t have a world without working to sustain it.
The world is in a climate emergency, and we live in the world. It’s just common sense to use what influence we have to promote green versions of widely-available technologies. We asked everyone on the team to investigate the cleanest options for all of our software and hardware needs, and will encourage our clients to do the same. This is just one way we intend to live this value, with ideas for products on the horizon that can help our enterprise customers optimize their models for sustainability, beyond just fairness and accuracy metrics alone.
We will act as agents of anti-oppression.
It’s not enough that the world be better for the elite few. Responsible work entails anti-oppressive work, requiring consistent reflection and thoughtful engagement with feedback and critique. We live in a global community, where oppression is no longer limited to hard geographical borders around the world. We commit to upholding democratic ideals above all, whether those are in danger from political or monopoly influences alike.
Only scientific rigor can generate reliable, repeatable results.
Great results only matter if they’re reproducible, and borne of sufficient experimentation and research. We therefore promise to pursue our research goals not with some prior-held belief of what the results should look like, but instead with an open mind about the possibilities that can be generated through exhaustive, careful work.
That’s all for now!
At the end of the week, it was hard to say goodbye to our temporary home full of such wonderful people. But we know we’ll see each other again soon, when the stars align and our team is much larger than it is right now.
Check back in to follow us on our madcap startup journey, where we hope soon to reveal the core functionality of our platform in the weeks ahead. If you’re interested in giving our software a try, feel free to reach out to us via info [at] getparity [dot] ai.
