Great new collaborative tool to assess the impact of LLMs. impactbench.media.mit.edu

Why are bridging conversations effective at reducing toxic polarization? One theory is that they break down our stereotypes by defying our expectations of what a typical counter-partisan believes. A new preprint - cleverly using chatbots - lends evidence that is the case. arxiv.org/pdf/2606.15901

I used to feel slightly lazy and needy each time I insisted on speaking to a *human* customer service agent. Now it feels like I'm doing my bit to preserve humanity.

I learned so much in this talk - about what users value in their subreddits, the spectrum of rules mods use and, most fascinatingly, what rules are connected to healthier communities. If you missed, this is a great write-up, but @galenweld.bsky.social is so fun to listen to it's worth the watch.

Someone studied the landscape of Bridge-building initiatives in the U.S., including @listenfirstproject.bsky.social's coalition of 400+ organizations, and created a map.

I love a research review. This one looks at studies of interventions designed to mitigate the negative effects of digitally-modified images on teen and adult body-image. www.sciencedirect.com/science/arti...

Prediction markets - Kalshi and Polymarket - now have their money (just slightly) on the Dems taking the Senate. That was a far-fetched dream six months ago.

The most dangerous AI on record may get into the hands most inept administration in history. If you weren't already terrified enough. www.wsj.com/tech/ai/anth...

Love a review article. This one is of studies from 2020-2025 looking at AI's ability to detect cyber-bullying. link.springer.com/article/10.1...

A positive sign from new study: In survey experiments, participants were less likely to engage with content when they knew it was AI generated. link.springer.com/article/10.1...

Just learned about this inspiring letter and vision for a New Civic Future - founded on agency, connection and empowered communities. (And not surprised to see @thevillagesquare.bsky.social's Liz Joyner as an author.) Gives me hope that we're witnessing a civic renewal. www.newcivicfuture.org/letter

We've reached the point when more than twice as many Americans Strongly Disapprove of Trump as Strongly Approve.

I'm so grateful to people like Erica Dhawan for articulating why I can't bring myself to use ChapGPT. ericadhawan.substack.com/p/your-brain...

@juliaminson.bsky.social is one of the most thoughtful and passionate scholars I've been fortunate to meet - so I assume her book will be equally smart and on point. On a topic dear to my heart. Looking forward to my copy arriving in the mail. disagreeingbetter.com/order/the-book

There are more global users of Whatsapp than Facebook, but fewer than 5% of studies on digital media and politics looked at Whatsapp. A new paper makes the case for why that needs to change. www.ojcmt.net/article/soci...

New study & promising approach to *keeping* college students open-minded. www.tandfonline.com/doi/pdf/10.1...

Someone did a mapping of the US Bridgebuilding field. Cool. @listenfirstproject.bsky.social @civichealthproject.bsky.social onlinelibrary.wiley.com/doi/pdf/10.1...

In this systematic review and meta-analysis of up to 153 longitudinal studies: Social media use was associated with higher depression, behavioral problems, self-injury, and substance use, and lower self-perception and academic achievement tinyurl.com/47z3anym

New study: Researchers in Germany tested a version of "correcting misperceptions" to see if it could mitigate hostile attitudes between the far left and far right. I could. onlinelibrary.wiley.com/doi/pdf/10.1...

New study: AI chatbots can be effective at reducing polarizing attitudes - especially when directed to exhibit receptiveness and active listening. www.researchgate.net/profile/Mark...

🔮 How can we empower online communities to design AI agents tailored to their unique needs and norms? In our #CHI2026 paper, we introduce #Botender, a system that enables collaborative design of AI agents through 🔥case-based provocation🔥

So thankful for federalism - and the NGA. A reminder that it may be the governors who get us through this mess. thehill.com/homenews/adm...

I have officially joined the Luddite community. buttondown.com/nycofftech

An AI-enhanced discussion tool (www.swaybeta.ai) is gaining traction across campuses and, according to professors using it their courses, helping students have more productive conversations on hard topics with those they disagree with. That's good news. www.chronicle.com/article/can-...

Great review from @emilykubin.bsky.social and @vaparker.bsky.social on the challenges of researching what effectively reduces polarization - including knowing if and when reducing polarization should be something we want to do. www.researchgate.net/profile/Emil...

Not to get too maudlin, but this old (great) song is feeling all too current.

An RCT across 120 villages in Bangladesh suggests that film can be a powerful way to increase inter-group trust - especially if viewed by folks who are socially centrally connected. academic.oup.com/qje/advance-...

What might make difficult conversations across division more fruitful? This study, using data from StoryCorps' One Small Step project, suggests pairing partners across generations:

I am one white person who needed to hear this - and am grateful for it. In case you're like me...

Good title and great quote: "Success requires leaning into the power of a shared social identity. ... Sometimes it means leaning into an “us vs. them” mentality, but more often it means building solidarity within a group while also expanding the sense of who “we” are." tinyurl.com/4ujerece

Social media has been implicated as a divisive global force — though scholars debate how much blame it really deserves. But a new review of 30 empirical studies finds it can also be used to reduce distrust and animosity across divides. scholarspace.manoa.hawaii.edu/server/api/c...

Another data point on why it's always a good idea to pause and wait for the full story. Headline last night: Man Shot in Chest in Midtown Manhattan, Police Looking for Suspect. Headline this morning... abc7ny.com/post/police-...

A reminder that Covid is not the only thing we should be testing - and staying home - for. 3,500 New Yorkers were hospitalized with the flu last week they counted - compared to 450 with Covid. And it doesn't look like we've hit the peak.

I don't know how I missed this special issue on "Connective Democracy" and social media. Some great work thinking about how social media divides - and can be designed to connect us even when we deeply disagree. journals.sagepub.com/doi/full/10....

I'm not saying that the latest BEA report on GDP shouldn't be trusted. But I am surprised that no one is saying that, given what happened at BLS and the fact that even the most well intended bureaucrats respond to incentives, perhaps they shouldn't entirely be? www.reuters.com/world/us/for...

Great - actionable - research from @engagingnews.bsky.social

An interesting reflection on what we get right and wrong in predicting the impact of technology. A newly published 1960s story from Tolkien imagined the pollution from cars would destroy humanity. That didn't happen; humans & regulation reined in that risk... lareviewofbooks.org/article/isen...

New on the blog: a conversation with Sabhanaz Rashid Diya asks, whose values are baked into “prosocial” tech design? From content moderation to the limits of exporting the DSA, it looks at why global defaults keep failing the Global Majority. www.prosocialdesign.org/blog/prosoci...

Important research question - Can we change ranking algorithms for better societal outcomes? - independently tested in the field. So much to love about this study.

When people complain about NYC - its crime & homelessness, its smell - I try to explain what a country club it is compared to the New York I grew up in in the 70s. From now on I'm going to urge them to watch this. It's not out yet, but from the trailer it nails it. www.youtube.com/watch?v=VZ94...

For any who think about both online child safety and polarization in the U.S. This report from More in Common shows strong cross-partisan support for digital protections - and suggests that it's an issue that help bridge divides. moreincommonus.com/publication/...

Any chances Cuomo had of winning today have got to be shot by the ballot design. Only determined Cuomo voters are going to put the work in to find his name in a sea of independents and may be forgiven for getting confused when they they think he's running on the Protect Animals ticket w/ Silwa.

Food banks are prepping for a surge in demand. Good time to donate to your local one. Here's NYC's: give.foodbanknyc.org/campaign/675...

Another RCT showing that, yes, conversations across division make a difference. link.springer.com/article/10.1...

America's cowboys are beginning to catch on... “The National Cattlemen’s Beef Association and its members cannot stand behind the President while he undercuts the future of family farmers and ranchers by importing Argentinian beef in an attempt to influence prices." www.ncba.org/news-media/n...

Morning thoughts, as I contemplate the longevity of my laptop: If we think about the possibility of AI "taking over", don't we have to consider the reality that the chips & computer hardware its build on is subject to decay?...

Thoughtful piece from Fukuyama on the unique role the internet has played in the global rise of populism - after a decade, he admits, of discounting that role. www.persuasion.community/p/its-the-in...

Few surprises here: 7/10 of the safest states are Blue. 8/10 of the least safe state are Red - including Florida (47) and Texas (48) wallethub.com/edu/safest-s...

13% of Americans now think polarization is our most important problem, while 15% of Americans think our most important problem is either Republicans or Democrats. We're divided about why we're divided.

AI says it wants to be our friend. New Yorkers aren't having it.