Tiago C. Peixoto

Studying where AI actually helps in science by modeling four stages of research. AI excels at screening vast possibilities, but human judgment remains key for new ideas, from @professor_ajay, John McHale, and Alexander Oettl nber.org/papers/w34953

How to guarantee #AI failure in public (and private) sector organizations: a field guide. #ICT4D #AI4Good linkedin.com/pulse/how-guar…

New paper of mine on #AI and #Democracy or, if it still exists, #AI4Good

An initiative for more taxes on the super wealthy is not a radical idea but a response to radical inequality in the world, says José Gilberto Scandiucci, Brazil’s Minister-Counsellor to the UN at the #FFD4. news.un.org/en/story/2025/…

🤖 🏛️ What does AI mean for government, beyond a few chatbots and a lot of hype? I’ve put some of the ideas bouncing around my head into a vision paper, laying out what “the agentic state” might look like. 🧵

Will #agentic #AI transform government and simplify bureaucracies? Great insights and foresight by @luukasilves with @participatory @OttVelsberg in @wef globalgovtechcentre.org/about-whitepap…

Where [oh where] are the compelling generative AI use cases in public-sector workflows? techpolicy.press/why-generative… A reality check by @participatory

that's from "The Geography of AI Adoption: New Usage Data Offers a Glimpse on Which Economies May See Greater Impact" by @lukesjordan @participatory @darkgreener mitgovlab.org/news/the-geogr…

🚨New research out today - The Geography of #GenAI Adoption: New @AnthropicAI Usage Data Offers a Glimpse on Which Economies May See Greater Impact tinyurl.com/laborgenAI

[Hmmm] "Traditionally constrained by limited corruption data—often available only through infrequent and expensive audits—scholars now can generate synthetic corruption measures at scale" voxdev.org/topic/methods-… @participatory @Dzinnbauer

Re-imagining African Unity in a Digitally Interdependent World research-portal.uu.nl/ws/portalfiles…

We're constantly told how countries are run by the elites. However, mapping the economic elite has been incredibly tough to do. But now, a new dataset compiled by 70+ researchers provides data on over 3,500 elites in 16 countries:

Going #digital. Transforming government through smart procurement. I have always believed that agile #procurement is a key policy policy lever to modernise #government. Great @wbg_gov post by @participatory on Romania's digital ecosystem. blogs.worldbank.org/en/digital-dev…

@essemmeppi Great - along with single domain plain language/content design is IMHO one of the most underrated actions in digital government. Is it an RCT by any chance? In any case, thanks for sharing!

@participatory Not automation – but INPS in Italy is using LLM to simplify the language of letters sent to people, with human validation needed to ensure legal correctness. A case study will be published shortly on OPSI’s case study library

Looking for compelling *gen* AI use cases in public service delivery - ideally beyond the usual RAG/retrieval tropes. Genuinely interested in automation of public workflows, not just smarter chatbots or info search. What’s actually interesting and working out there?

The Empty Chair: Using LLMs to Raise Missing Perspectives in Policy Deliberations arxiv.org/pdf/2503.13812

A Replica for our Democracies? On Using Digital Twins to Enhance Deliberative Democracy philarchive.org/archive/NOVARF…

In a study modeling 270 diabetic retinopathy screening scenarios in China, the most cost-effective approach was a "copilot" model (AI + human). An estimated health benefit of US$4.64 million per 100,000 people, compared to manual screening. #AI #HealthEcon arxiv.org/pdf/2503.20160

New @MIT/@OpenAI experiment and the paradox of AI companionship: voice-based chatbots initially reduce loneliness compared to text, but higher usage across formats correlates with increased loneliness, emotional dependence, and reduced human socialization. arxiv.org/pdf/2503.17473

New research uses LLMs to simulate online discussions & evaluate moderation strategies at scale. Study finds game-based approaches outperform traditional guidelines & smaller models create more diverse conversations. #AIdeliberation arxiv.org/pdf/2503.18858