Brian Ellsworth

The doctor is always in - as long as you just get there through a chat window. More than 70% of Brazilians used AI for medical advice last year. But the risks are real: chatbots absorb misinformation, and spotting symptoms isn't diagnosing a patient. 🎙️ Produced by @iqblanco

#ExpertsofTheAmericas | How is AI reshaping Latin America and the Caribbean? In this episode, Brian Ellsworth (@brianpablo10), communications advisor and AI public policy analyst, joins AALAC’s Maite Gonzalez Latorre to break down the evolution of AI and what it means for the region. ▶️Watch here: youtube.com/watch?v=YrIkeQ… #ACLatAm @AtlanticCouncil

So long, Sora. OpenAI is shutting down its AI video app. Fun, but not very useful. And expensive to run. The AI race is shifting—from spectacle to discipline. Produced by @iqblanco

People detox from alcohol. They detox from social media. Now some are talking about detoxing from AI. Young Brazilians say overreliance on tools like ChatGPT is making it harder to think and create, according to Folha. Is AI detox the new Brazilian fad diet?

Would you elect AI to represent you in Congress? Colombians had the chance when a blue-skinned avatar named Gaitana ran on a promise to follow instructions citizens gave to a chatbot. Voters chose humans instead. But don’t expect that to be the last AI candidate.

Should AI be put in charge of missile systems? This was at the center of a dispute between AI developers and the U.S. military. Anthropic drew red lines. OpenAI set fewer boundaries. But should tech companies even be deciding this? It seems like a job for Congress.

If AI knows too much about you, it feels creepy. If it knows nothing, it’s useless. Brazilian researchers modeled a world where regulation blocks AI personalization. Result? 🇧🇷 GDP 1.6% lower over a decade — ~$160B lost. How much should AI be allowed to learn? New video 👇

Can AI make sparks fly? I asked my chatbot alter ego to try his hand at seduction. It did not go well. A new video from LatAm Prompt 🎬 Love is in the AIr. Produced by @iqblanco

Super Bowl ads? Yes. 🍿 Ads in my chatbot? No thanks. New LatAm Prompt video on what changes when ads move into AI. With help from El BrIAn & @iqblanco

Texas dreams are big, and Venezuela is once again as hot as the mythical El Dorado. --> fun one with @ArathySom --> US oil capital Houston buzzes as industry limbers up for Venezuela oil rush reuters.com/business/energ…

For years, AI improved by scaling: more data, more compute, more money. That strategy is hitting limits. More scaling may not get us better videos of Will Smith eating spaghetti. What comes next? Check out my latest video for LatAm Prompt.

AI data centers can draw as much power as a steel mill — but they can spike demand instantly. Yet grids still treat them as commercial users, like malls. For fragile Latin American systems, that could mean higher prices or more blackouts. Produced by @iqblanco

Lawyers using AI to draft filings? That’s so 2023. Now a judge in Argentina is in trouble — after his ruling included: “Here’s Section IV ready to copy and paste.” The court tossed the conviction. AI isn’t just passing the bar. It’s writing the verdict. Produced by @iqblanco

Could AI make me Venezuelan for 30 seconds? Who knows ... but if AI doesn't come through for you, you can always count on El BrIAn 😁 With help from @iqblanco

Had a blast at the StartSe AI Festival in São Paulo 🇧🇷 Workshops on vibecoding, context engineering, and the latest AI tools. Brazil’s AI story is being built from the ground up. A reminder that it’s not the tech, but the people, who make AI real. #AI #Brazil #StartSe #LatAm

AI isn’t just scraping the news — it’s reselling it. In Brazil, new services use AI to rewrite media stories and sell them as custom feeds — sometimes even with real news logos. How can media respond to AI scraping? Check out this week's video.

AI developers rarely have to explain how their models get this close to artists’ work. Brazil wants to change that — with a law that flips the burden of proof. Creators wouldn’t have to prove they were copied. Developers would have to prove they weren’t. 🎥 Produced by @iqblanco

Can AI use pirated literature? No. Anthropic paid $1.5B to settle a dispute over the use of pirated novels by writers including Brazil’s Clarice Lispector. But a judge ruled AI systems can train on novels without author permission. It's complicated ... Produced by @iqblanco

One of the best tools in a writer’s kit has been tarnished with the stench of AI slop. Writers in Brazil say they spend hours rewriting text just to avoid being flagged — sometimes making it worse. Why do chatbots love the dash? My theories in this new video with @quijada_ivan.

Read LatAm Prompt's coverage of the impact of AI onn Mexico's dubbing artists . latamprompt.substack.com/p/whose-voice-…

Who owns voice in the AI era? Mexican dubbing artists are asking this. Already facing replacement by AI voiceovers, they were outraged when their government used an AI clone of renowned voice actor Pepe Lavat without permission. Produced by @ivanqblanco