As many entities both in and out of government call for scaled-up and AI-enabled public participation, I hope they will bear these distinctions in mind. Much more in the @BostonReview piece itself, and happy to hear any thoughts as this is part of a larger ongoing book project.

If officials or other audiences want to be able to learn both novel and specific information from public inputs, then there is no substitute for human context-specific knowledge. Someone needs to actually read what the public says, and this can be expensive in time and cost!

AI can also automate classification of individual inputs on the basis of already-known examples or rules, but this approach is limited by potential algorithmic biases, and inability to detect novel forms of importance that were unrecognized ex ante (and so not in training data).

Whether or not AI can be useful depends on what we actually want to learn. It can help aggregate public inputs into numeric averages or thematic summaries. But sometimes we want to learn specific problems, proposals, or perspectives, not aggregated preferences or themes.

When civic engagement operates at mass scale -- like a consultation, complaint, or proposal platform that receives tens of thousands of contributions -- it creates an information overload problem for government officials and other potential audiences. Can AI help?

There are many promising applications of AI (and similar tools) for participatory and deliberative democratic innovations, such as for improving accessibility, inclusion, moderation, or outreach. But there are other uses being discussed for which we need to be more cautious.

Very happy to share this essay in @BostonReview on the limits of AI for civic engagement. A brief thread:

The worst attack on 'tenure' at the London School of Economics in years was just revealed by the @ucu @LSE_UCU branch. It is ongoing and impacts all LSE academics, from Research Staff to Full Professors. Please share widely, as we're in trouble here (1/12)

Excellent text by @berliner_dan, offering a sobering account of #AI’s potential to transform democratic processes.

Many are hailing the promise of digital technologies like new AI tools to enhance civic engagement and improve democracy. Daniel Berliner scrutinizes this optimism, explaining what AI can’t do for democracy: https://t.co/0Dzf8zVIdy

New working paper with @NikKalyanpur and @berliner_dan‼️ Activists use transparency ("leaks") to try to dismantle the opaque offshore world. Does this work? We identify a new effect of offshore leaks: they encourage stronger private governance in the financial sector. 1/

This is NOT what America is about. America is about *opens history book* uh oh *Frantically starts flipping though pages* uh oh. oh no. no no no. uh oh

We’re #hiring a Communications Program Officer to help lead storytelling at OGP! If you have 5+ years in #communications, strong writing/editing skills, and experience in #opengov and #nonprofits, apply now! 📍 Washington D.C., Belgium, London 👉 https://t.co/I3bp1NJp0J

We at @acdatacollectiv have a brand new report with @anticorruption about how deep the conflicts of interest go at this year's #COP29 in Baku, AZ The UN climate summit is at real risk of : - corruption - capture by fossil fuel companies - greenwashing https://t.co/KEa2x1x0k5

Citizens hear about democratic backsliding on the news, but they meet it in interactions with street-level bureaucrats. A @Gov_Journal SI, co-edited with @NadineRaaphorst and @gabilotta, explores how populism and democratic erosion shape the work of the frontline public sector🧵

New on Early View: Notice the comment? Chinese government responsiveness to public participation in the policymaking process https://t.co/X5mWW3VyBW

“By adopting a cross-chronological and world-historical approach, the book challenges the abiding association of bureaucratic rationality with 'modernity' and the so-called 'Rise of the West'.”

📢 Job alert! Come join @LSEMethodology! We're a friendly interdisciplinary dept, looking for a new colleague with expertise in qualitative research methods. (No need to be a 'methodologist' - just someone who cares about doing careful, considered work). Feel free to reach out!

There is no better way to understand how shallow support for democracy is in the UK than to ask 'should hereditary aristocrats have any say in governing the country in the 21st century' and see what response you get.

New research alert! 🚨 My latest article, now online at @LSQjournal , leverages original data on anticorruption initiatives and subnational variation in term limits to show how reelection incentivized responsiveness to anticorruption demands.