Daniel Trielli

@danieltrielli.bsky.social

Brasileiro. Inconstant poster in English and Portuguese. Assistant Professor of Media and Democracy at the University of Maryland's Philip Merrill College of Journalism. https://danieltrielli.com

Momento 1: Vejo a notícia que o Gov Br tá fora do ar; Momento 2: A notícia me faz lembrar que preciso checar um negócio no Gov Br; Momento 3: Tento acessar o Gov Br; Momento 4: O Gov Br não abre; Momento 5: Penso "Que estranho, será que tá fora do ar?"

It's amazing how often Brazil has great examples of how wonderfully weird it is. Yesterday, someone changed the election registration of a presidential candidate to list him under a different party. As a prank, apparently? By tomorrow we'll have forgotten about it. Something weirder will happen.

Nunes Marques cancela filiação de Flávio Bolsonaro ao Missão e mantém senador no PL

Flávio Bolsonaro teve registro de candidatura à Presidência travado após filiação ao Missão aparecer na Justiça Eleitoral

estadao.com.br

Conference travel is never easy for me. I always get this nagging feeling on the plane that I’m traveling hundreds of miles an hour in the wrong direction, away from my family. But I’m still a little buoyed by the conversations and presentations I saw at AEJMC this year.

I made this exact point to my students in two separate classes this week: “Hallucination” is barely useful as a metaphor. It is a human experience that implies a warping perception of reality. AI systems have no perception of reality; they store information about language representations of it.

Nash@radiodeadair.com · 5mo ago

The reason you'll never be able to stop LLMs from "hallucinating" is because they don't hallucinate. They output pattern-matched responses, things that appear similar to answers based on a database. It can't know what's in that answer because it doesn't know anything.

Specifically, the idea of source credibility just doesn't apply straightforwardly to LLM interactions. When I get info from another person I can ask: how does the content relate to their expertise? Their lived experience? Their interests and biases? 8/

🎶 It’s the most wonderful time of the year… It's December, and that means that it’s Nieman Journalism Lab predictions season! I really like these. It doesn't matter which are proven right or wrong; this is a chance to see what a bunch of news media thinkers are worried or excited about.

Screengrab of webpage that reads:

"NiemanLab

Predictions for Journalism 2026

Each year, we ask some of the smartest people in digital media what they think is coming in the year ahead. Here's what they told us."

I’m happy to share that I have a chapter in the about-to-be-published Handbook of Artificial Intelligence and Journalism (Wiley Blackwell)! The book, edited by Aynur Sarısakaloğlu and Martin Löffelholz, is an essential read if you’re working on AI in newsrooms or journalistic transformation.

Bild

I haven't been posting in a while because everything I want to post is either unadvisable or irrelevant. But now I want to announce that, in the midst of all this, against all sense, logic, and practicality, we have adopted a dog. And it's the best thing we've done in a while. Meet Princess Daisy.

A white and brown dog is lying on the couch, looking at the camera.

Our 4yo: "I want to watch the movie of the unicorn that grows big and turns into a T-Rex." Me: "Ok, is that a real movie, a movie you're developing right now, a game you played, or a dream you had? Because if it's a real movie, I have follow-up questions to track it down."