AI, Tell Me A Story

9 Things I’ve Learned in My First Year Working at the Intersection of AI and Communications.

Photo by Alex Knight on Unsplash

Today marks my first year at the Partnership on AI (PAI), a nonprofit founded to ensure that artificial intelligence (AI) benefits people and society.

Given that my role — partner and internal communications — was the first of its kind for PAI, and that I come from a nontechnical background, I thought I’d share some lessons I’ve learned doing comms in a new-to-me field.

I imagine this will be most helpful if you:

• are new to AI (in any role),
• are currently working in comms in a technical and/or information dense field, or
• are starting a comms role in a new-to-you subject area.

These lessons are split into three sections, so you can take what’s helpful and leave the rest:

1. The Knowledge — how to learn what (you think) you should know
2. The Process — how to work without playing yourself
3. The People — who to work with and learn from

The Knowledge

Photo by Christin Hume on Unsplash

LESSON ONE. Steep yourself in subject matter.

There’s a not-always-harmonious chorus of voices and perspectives on artificial intelligence these days, so much so that it can feel overwhelming to know where to begin as a non-technologist.

Up front, I’d suggest the best place to start would be to take a deep breath, dive in blindly, and trust that you’ll pick up the jargon and concepts you actually need along the way. Before you panic, I promise you that familiarity will come faster than you think!

In my first week at PAI, I produced social content for a paper on facial recognition technology, which I came in knowing diddly squat about. In my first month, I edited an academic paper on reconciling legal and technical approaches to algorithmic bias. Was I in over my head? Absolutely! Did I learn a ton, and quickly? Heck yeah. Did I spend a lot of time pacing back and forth in my apartment, mumbling concepts on algorithmic fairness out loud to myself whilst willing my brain cells to figure it out already? You betcha.

Assuming you fall into one of the three buckets noted above, you’re precisely in this role because you’re NOT an expert on all things (AI or otherwise), so don’t make the mistake of holding yourself to that standard on Day 1. No matter how intimidatingly intelligent your coworkers might sound, you hear me? That said, nobody likes to dwell in their own ignorance, so here are a few favorite resources that I used to learn about AI:

Newsletters:

• The Algorithm from MIT Tech Review (subscription only)
• Import AI
• The Creative AI Newsletter

Books & Other Reads:

• The A-Z of AI from The Oxford Internet Institute and Google
• Anatomy of an AI System by Kate Crawford and Vladan Joler
• A People’s Guide to AI from Allied Media Projects ($7)
• You Look Like A Thing And I Love You by Janelle Shane ($28)
• Race After Technology by Ruha Benjamin ($20)

Podcasts:

• Women in AI from RE•WORK
• Radical AI
• Rabbit Hole from The New York Times
• Surveillance And The City

(Virtual) Events & Workshops:

• AI for Anyone Workshops
• Stanford AI for Social Good Workshop
• Data & Society Events
• Karen Hao’s AI Talk series

Note: remember, the idea here isn’t specific to AI; no matter your field, there are likely similar resources in these formats available.

How did I find resources to learn a new subject quickly?

1. I asked coworkers for recommendations on their favorite sources. My coworker and fellow productivity junkie Rosie Campbell had previously written about her favorite AI newsletters, and our PR wizard Peter Lo turned me on to Karen Hao’s amazing work.
2. I followed the work of organizations we already worked closely with. This led me to the events put on by Data and Society (a PAI Partner), at which I learned of the cool research of folks like Madeleine Clare Elish.
3. I did some typical internet sleuthing based on my own interests. This is how I discovered Black in AI, Queer in AI, and AIxDesign, interdisciplinary communities that explore AI and its many topical and demographic intersections.
4. I followed people who, from the above, seemed interesting. This is where I fell into many a Twitter rabbit hole. I highly recommend following the AI work of: Timnit Gebru, Meg Mitchell, Ruha Benjamin, Emily Bender, Alex Hanna, Devin Guillory, Safiya Noble, Rumman Chowdhury, Rachel Thomas, Ali Alkhatib, Jordan Harrod, Meredith Whittaker, Hanna Wallach, Deb Raji, Miles Brundage, Umang Bhatt, Alice Xiang, Jingying Yang, Tina Park, Jeff Brown, Claire Wardle, Emily Saltz, and Claire Leibowicz.

LESSON TWO. Source + follow trusted voices.

From my PR background (glad those days are behind me 😅), I assumed there’d be a few well-respected journalists covering AI who I could read regularly and trust to keep me up to date.

To avoid information fatigue, I focused on finding journos who:

1. wrote for publications that were industry-specific but not too niche
2. approached stories with accessibility and nuance rather than one-dimensional “AI is evil” or “AI is best thing since sliced bread” perspectives
3. (bonus) ideally were people of color, given how overwhelmingly homogenous 🙄 AI can be as a field

From that criteria, I’d recommend following these folks if you’re looking to learn more about AI:

☞Karen Hao, Senior Reporter at MIT Technology Review
☞Khari Johnson, Senior Staff Writer at VentureBeat
☞Jack Clark, Import AI newsletter author
☞Casey Newton, The Platformer newsletter author

Coming from prior comms roles in information-heavy fields — labor policy, immigration law and policy, and higher ed access — I can attest that lessons #1 and #2 are tried and true steps to speed up your learning journey (and enjoy nerding out, if that’s your thing).

LESSON THREE. Prioritize accessibility of your content.

Now that you’re gaining more familiarity with AI/your subject matter of choice, don’t pull the ladder up behind you.

If you’re in a content-producing comms role, use your own experience as a beginner to build empathy towards your future readers who may also not be elbows-deep in the minute details of your field.

Spell out acronyms. Provide simple definitions each time you introduce key concepts. Be succinct. Tell the reader what you want them to do or know; don’t force them to use deductive reasoning as an intellectual thought exercise just to flex how disruptive/innovative/fancy you are! Use different formats — graphics, audio, video — to make your work digestible for multiple platforms and for different learning styles. Add captions, alt text, and transcriptions where possible.

Always know your audience, of course; carry on if you’re talking directly to theory-hungry senior researchers and no one else. But I’ve found that nonprofits in information-dense fields — those in technology, law, or policy especially — often prioritize expertise and intellectual rigor of content over accessibility to the reader, a short-sighted mistake that can narrow your audience instead of broadening it.

The Process

Photo by Kelly Sikkema on Unsplash

LESSON FOUR. Structure begets sanity.

Real talk: just because you CAN be scrappy/thrive under chaos/grind on deadline doesn’t mean you SHOULD (I repeat this, ad nauseam, to myself constantly).

In nearly all of my past comms roles, I’ve been 1) the first comms hire for the org or the first in the function and 2) been a one-person team. I love building from scratch and happen to find infrastructure work fun, but I 10/10 would not recommend (I kid, kinda).

All that to say, prep yourself before you wreck yourself, I implore you. Here are three things that have helped me tap into the magic of process over the last year:

• An editorial calendar. As our Comms Team grew from one to five folks (I was #2) and we had the capacity to pump out more content, I saw a need to organize ourselves internally as a team to make hitting publish a less stressful endeavor. I’m an Asana fan girl from my time at my former company Magoosh, so I turned to Asana’s visual board feature to build us a calendar. It gives us an at-a-glance sense of our workload as a team at any given time, an archive of our past work, and a space to take a breath and reflect each month with a retrospective of our content.

• A staff-facing comms request form. Once we hit our stride using the editorial calendar, we realized it’d be more efficient to have requests from staff come to us rather than wade out on a content-finding mission each week. I built an intake form in the same Asana board that gave us a pretty seamless way to collect the details we’d need to decide whether to produce a piece of content. Staff fill out the form when they need something, it’s automatically added to our editorial calendar in an “incoming requests” column, and we review all requests weekly during our team meetings + aim to respond to folks quickly.

• A staff-facing media log. I’ve used some version of the same Google spreadsheet in my last few comms roles to track press coverage. While I love tools like Muck Rack that come with built-in reporting dashboards, I appreciate the presence of mind that comes from maintaining an in-house log as well. It makes pulling content for newsletters, blog posts, and annual/funder reports a breeze, enables you to pattern-spot across frequent topics or spokespeople, and can give you content ideas when your creative juices are parched.
  _
  My go-to points to track for press coverage: date published, headline, URL, outlet, journalist name, topic area, type of coverage (resource citation, quote, feature, etc), sentiment of coverage (positive, negative, neutral), spokespeople mentioned, and text of any spokesperson quotes. For smaller orgs without huge volumes of coverage, this isn’t as tedious as it seems and makes for excellent maintenance work for someone more junior if available.

Tidying Up With Marie Kondo, Netflix

LESSON FIVE. Get your house in order.

While we’re on the topic of process, if you’re in a role that provides the autonomy or access needed to advocate for broader, org-wide structural improvements, doing so could pay dividends for your own work.

For me, this looked like pitching our Comms and Executive Directors on adding internal comms to my role as a new function for the org. As we’re just a few years past our founding as an organization, we were experiencing a lot of telltale growing pains — lack of coordination, communication silos and misunderstanding, decision-making frustrations, etc. Drawing upon what we’d learned from testing out more structure and process as a Comms Team, I set out to figure out where we’d get the most bang for my half-time buck of internal comms.

I quickly settled on our weekly All Hands meeting as Important + Urgent low-hanging fruit. Working in collaboration with HR & Operations, we’ve spent the past five weeks rolling out a beta version of a new All Hands structure, and have happily settled on A New Path™. We moved away from our previous model of all-verbal, share-if-you-want updates on an ever-revolving list of topics, with occasional 45 minute presentations sprinkled in. We found this favored folks with certain personalities and speaking styles over others, made it impossible for those who were unable to attend to keep up, and made it challenging to follow the conversation around next steps on major decisions.

We now work from a standard deck template that folks fill out a couple of days before each All Hands; each department gets one slide and picks one representative to give a short, top 3 highlights update. We open the meeting with a rotating Zoom DJ who plays one song & asks a check-in question as folks get settled, and encourage folks to volunteer to facilitate the whole meeting to ease the power dynamic that can come from nonstop airtime from leadership. We spend the second half of the meeting on 15–20 minute “Deep Dive” presentations from across the org, with time for Q&A afterwards. For now this is working well for us, but it’s totally possible that we’ll switch gears when our growth hits another inflection point.

The People

Photo by Brooke Cagle on Unsplash

LESSON SIX. Connect dots across disciplines.

For better or worse, my brain is happiest when it’s exploring and learning at the intersection of disciplines, and AI has been no exception. Some AI intersections I’d recommend learning more about:

1. AI x Criminal Justice: The rising use of AI and other technologies in the carceral context is especially troubling given the longstanding history of racial inequities and civil/human rights challenges. Learn how the #NoTechForICE campaign from Mijente is fighting back against tech used to surveil and target immigrants, and how the U.S. Federal Bureau of Prisons has continued to use faulty algorithmic risk assessment tools to make eligibility determinations around pretrial detention and home confinement.
2. AI x Art: I don’t remember how I first learned of the work of transmedia artist Stephanie Dinkins, but it led me to this cool Timeline of AI Art from AIArtists.org. Stephanie is also behind AI Assembly, a collective grappling with the question of what algorithmic systems demand from us. Learn about the work of other AI artists like Joy Buolamwini and Mimi Onuhoa, and check out this field guide to making AI art responsibly from PAI.
3. AI x Fake News: The Trump administration was sadly and maddeningly a perfect storm of misinformation, media manipulation, and an overall erosion of public trust in journalism as an institution. Through the work of my amazing AI & Media Integrity coworkers Claire and Emily, and resources from PAI Partner organization First Draft, my eyes have been opened to how mis- and disinformation have become permanent fixtures in our daily news and social consumption.
   _
   This, in part, inspired me to volunteer with Common Cause as an Election Protection Volunteer in Arizona on Election Day. Along with many other folks passionate about voting access, I monitored day-of Twitter discourse from voters, sharing voter support resources and reporting/intervening in suspected misinformation in real-time to ensure and uphold the integrity of our voting process.
4. AI x Law: I’ve learned so much about the potential legal shortcomings of the common technical approaches that are used to mitigate bias in algorithms from my former (stylish!) coworker Alice Xiang. Essentially, although it’s great that many folks are now paying attention to and working to reduce or eliminate biased outcomes from algorithms, it’s important that those mitigation approaches don’t wade into questionable legal territory by brushing against the limits of anti-discrimination law. A heady topic that I’ve grossly oversimplified, to be sure, but a fascinating one that you should definitely learn more about.
5. AI x Policy: Similarly, I’ve enjoyed tapping into my policy wonk tendencies through learning about AI policy from my former and brilliant coworker Lisa Dyer. From Lisa I’ve learned that policymakers are hungry for better understanding of facial recognition and other AI technologies, and that immigration laws — through their arduous visa requirements — can inadvertently squash budding AI talent.

Art by Libby VanderPloeg

LESSON SEVEN. Listen to + make room for those not invited.

I personally believe that in order to further PAI’s mission, which is to “bring diverse voices together so developments in AI advance positive outcomes for people and society,” we must always return to how people — real people and communities, not just the abstract idea of them — are affected or harmed by AI technology.

Having worked in immigrant rights for the past decade, I’m well aware of how power dynamics in nonprofit work limit the ability of directly impacted communities to have a say in policy and regulatory decisions that have often dire consequences for their quality of life. Even then, working in AI has opened my eyes to the vast and unchecked power that private technology companies yield over our daily actions, particularly the invisible power that drives decisions behind the scenes around how and for whom technology is developed.

My talented coworkers Tina and Jeff have introduced me to some important concepts geared at making room at the AI table. One of these ideas involves moving beyond traditional participatory design approaches towards less extractive methods for inclusion in the technology development and deployment process; you can learn more here and/or sign up to share your experience doing similar work in the AI/ML space. The other idea examines why efforts at diversifying the faces working on AI have failed, leading to high attrition rates in the field for people from non-White, non-male backgrounds; if that happens to be you, you can also chime in with your thoughts on this study.

LESSON EIGHT. Resist role blinders.

Though your role scope may be limited to comms or something else, don’t be afraid to stretch beyond your immediate role when the opportunity presents itself, workload and mental energy withstanding. Like athletes in off-season, working across functions is a great way to accelerate your learning through professional “cross-training.”

For me, over the past year this has looked like serving on hiring committees for roles spanning Research and Comms across the company. Though I’m not a people manager, interviewing 15 candidates over three hiring cycles helped me hone my own narrative about (and understanding of) our work, taught me how to ask more thoughtful and less biased questions, and reminded me once again of the importance of tapping into the beginner’s mindset. I’ve also been a notetaker for a user research study run by one of our fellows, a co-facilitator for a workshop we ran, and have sat in on insightful conversations with my researcher coworkers.

Above All Else

Photo by Tachina Lee on Unsplash

LESSON NINE. Aim for Curiosity > Mastery.

As communications professionals, we often forget that our role isn’t to be the subject matter expert in the room (though many of us get there with time just through sheer exposure and immersion). Our work is that of an expert translator, persistent bridge builder, and graceful connector of dots, weaving meaning and understanding from complexity and detail.

“Research is formalized curiosity. It is poking and prying with a purpose.”
― Zora Neale Hurston

No matter your role or the extent of your subject matter expertise, I hope these lessons inspire you to both cut yourself some slack as you learn AND to nurture the flame of curiosity that drives your learning.