Lightnews — Scholar-powered news

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Jul 19

Latin LLMs: Buy, Build or Fine-Tune?

LatamGPT aims at a regional language model trained in regional data at Chile’s solar-powered supercomputing center.

Switzerland’s own in-house LLM shows that sovereign AI is possible—but is it the best path for Latin America?

www.linkedin.com/pulse/latin-...

Latin America's search for AI sovereignty

From the foothills of the Andes, a group of Chilean engineers is building an ambitious AI experiment: LatamGPT, a regional language model trained not in California, but at Chile’s solar-powered superc...

www.linkedin.com

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Jun 30

"Latin America faces a choice that will determine its economic future: develop AI governance on its own terms, or become a regulatory colony of Silicon Valley and Brussels."

Escribí sobre cómo pensar la regulación de la IA en América Latina.
americasquarterly.org/article/regu...

Regulating AI on Latin America’s Terms

The region has a narrow window to develop its own approach, taking inspiration from international frameworks while adapting them to local needs.

americasquarterly.org

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Jun 16

"Credentials say what you know, employment records show what you’ve done. If we can’t track experience, we can’t recognize it and reward it. And experience will soon be our last chance to complement the algorithm."
My new piece for @brookings.edu.
www.brookings.edu/articles/emp...

Employment records: The missing piece in the US labor market?

Eduardo Levy Yeyati discusses employment records systems and explains how they could potentially transform the U.S. labor market.

www.brookings.edu

1

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Jun 13

“We have public education, health, and transportation policies, but we don’t have public digital connectivity policies. That is absolutely inexplicable today.”

Chile-backed model Latam-GPT plans to make its public debut in September. foreignpolicy.com/2025/06/13/l...

Where Does Latin America Stand in the Global AI Race?

Chile-backed model Latam-GPT plans to make its public debut in September.

foreignpolicy.com

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Jun 2

On Argentina's stabilization and growth, and automation & the future of work (in Spanish).
www.ambito.com/economia/edu...

Eduardo Levy Yeyati: “Una fuente de riesgo es que en 2017 parecíamos haber salido de la crisis, pero duró poco“”

El economista analiza las condiciones actuales para atraer inversión productiva. Resalta la estabilidad del dólar y el resultado electoral como requisitos clave para invertir. También advierte por la ...

www.ambito.com

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · May 22

Here, the complete column:
www.brookings.edu/articles/sma...

Smart AI regulation strategies for Latin American policymakers

Eduardo Levy Yeyati lays out essential steps for Latin America to adapt and create smart, inclusive, and forward-looking AI regulation.

www.brookings.edu

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · May 22

With fewer legacy systems, growing tech ecosystems, and a strong tradition of social rights-based regulation, Latin America is a natural sandbox for forward-looking AI governance. In a recent piece, I lay out the key ingredients and essential steps for smart AI regulation in LAC.

1 1

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Apr 21

www.infobae.com/educacion/20...

Eduardo Levy Yeyati: “La inteligencia artificial pone en crisis la centralidad del trabajo en nuestra sociedad”

En un encuentro organizado por Ticmas y el Grupo de Fundaciones y Empresas (GDFE), el economista conversó sobre el impacto de la inteligencia artificial en el trabajo, la educación y la estructura soc...

www.infobae.com

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Apr 14

We wrote with Juan Carlos Hallak for Americas Quarterly about Latin America's failed experiment with import substitution industrialization (ISI), and the lessons that can be drawn for the U.S. tariff war.
americasquarterly.org/article/fail...

Failed Protectionism: What Latin America Can Teach Us

The region’s failed experiment with import substitution industrialization (ISI) offers a stark warning: embracing it may breed inefficiency, corruption, and stagnation.

americasquarterly.org

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 24

Una lección de las corridas de los 80-90 fue que una flotación sucia es preferible (más resiliente) a una banda porque no ofrece al especulador un blanco (el techo de la banda) que atacar. De ahí la prevalencia de los regímenes flexibles en el nuevo milenio.
Fuente: ideas.repec.org/p/sad/wpaper...

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 23

Gen AI offers an opportunity for education—not to replace teachers, but to support them. In this study we show how AI can act as a scaffold—not a shortcut—for learning, and why LAC is well-positioned to lead the integration of AI in the classroom.
Link (English & español): utdt.edu/ver_contenid...

1 1 3

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 21

"Digital advertising excels at predicting cultural fit. [...] If these factors can be used to sell cars, it is hardly a leap to imagine they could also help predict job fit."

My piece on alternative job scoring for @brookings.edu.

www.brookings.edu/articles/dig...

Digital footprints and job matching: The new frontier of AI-driven hiring

Eduardo Levy Yeyati and Ian Seyal discuss how AI can be responsibly leveraged to assess the more qualitative attributes important to hiring, like skills and cultural fit.

www.brookings.edu

1

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 19

Our paper on What Works for Active Labor Market Policies is already available online.

Ya está disponible online nuestro artículo sobre qué funciona en politicas pro empleo, donde comparamos evaluacion de experiencias recientes en el mundo.

Link: economia.lse.ac.uk/articles/10....

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 18

History teaches that harder than containing inflation with an exchange rate anchor is exiting the anchor in time. When the moment comes, nobody wants to change a winning strategy; once the moment´s passed, there is no easy way out.

"You can check out any time you like, but you can never leave."

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 17

La confusión entre democracia y liberalismo nubla el debate sobre la nueva derecha, que es a la vez democrática (popular, incluso populista) y antiliberal (aversa a la libertad de expresión y a la separación de poderes). Su oposición natural no sería la izquierda iliberal sino un liberalismo amplio.

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 16

Acá, el cuento. Muy meta, un tanto académico (muy creative writing workshop americano, sobre todo en la elección de palabras e imágenes), muy "profesional".
¿Creación o elaborada copia?
www.theguardian.com/books/2025/m...

‘A machine-shaped hand’: Read a story from OpenAI’s new creative writing model

Posting the piece online, OpenAI’s Sam Altman said this is the first time he has been “really struck” by AI writing

www.theguardian.com

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Mar 16

Escritores britanicos sobre el cuento corto hecho por el modelo de escritura creativa de Open AI: “un elegante vacío” (Nick Harkaway); “genuinamente inteligente” (David Baddiel); “hermoso y emotivo” (Jeanette Winterson).
www.theguardian.com/books/2025/m...

‘OpenAI’s metafictional short story about grief is beautiful and moving’ | Jeanette Winterson

I think of AI as alternative intelligence – and its capacity to be ‘other’ is just what the human race needs

www.theguardian.com

1 1

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Feb 19

"DeepSeek suggests the future may be defined not by a single dominant technology but by a range of competing standards. Regardless, LatAm must prevent swapping one form of dependence for another."

My piece on the implication of DeepSeek for LatAm's AI agenda:
americasquarterly.org/article/deep...

DeepSeek Reveals Latin America’s AI Crossroads

China’s new large language model, LLM, offers the region a chance to seize AI on its own terms but also brings risks of dependency.

americasquarterly.org

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Feb 17

Esta nota sobre la pelea de Milei con la profresión explica las comillas del tuit anterior.
www.lanacion.com.ar/politica/mal...

Malestar entre los economistas por la acusación de Fantino a un think tank

El periodista aseguró que miembros del CEDES “necesitan la devaluación” porque “cobran en dólares” y varios influencers libertarios reprodujeron esa misma información; sus declaraciones fueron repudia...

www.lanacion.com.ar

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Feb 17

[ARCHIVO] Hace un par de años escribimos sobre el mundo cripto con un querido amigo y distinguido "econochanta".
#libra #milei
www.brookings.edu/articles/cry...

Cryptomania: The good, the bad, and the ugly

Eduardo Levy Yeyati and Sebastian Katz critique the crypto revolution and the advent of central bank digital currencies.

www.brookings.edu

1 1

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Dec 21

Se la va a extrañar. QEPD.
www.youtube.com/watch?v=UeGE...

Beatriz Sarlo, con Levy Yeyati: “A los políticos los juzgo viendo cómo les fue en su territorio”

YouTube video by Todo Noticias

www.youtube.com

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Dec 21

Mi nota sobre el rol del conocimiento práctico en la complementación entre humanos y algoritmos para el blog de at @brookings.edu.

My piece on the role of experiential knowledge in strengthening the complementarity between human workers and algorithms.

www.brookings.edu/articles/why...

Why gen AI can’t fully replace us (for now)

www.brookings.edu

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Nov 29

Cómo regular el cambio tecnológico en su fase exponencial sin trabarlo o mirarlo por el retrovisor?
Para orientar el debate escribimos con Angeles Cortesi esta guía introductoria sobre regulación de IA en LAC (versión en español pronto en este mismo link):
utdt.edu/ver_contenid...

Translate post

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Nov 25

Hay un meme circulando donde alguien pregunta a ChatGPT qué puede decirle de sí mismo y el algo le da una descripción "profunda" de su personalidad. No te dejes engañar pensando que hay algo profundo ahí: es sólo un algoritmo actuando como un horóscopo.
simonwillison.net/2024/Oct/15/...

ChatGPT will happily write you a thinly disguised horoscope

There’s a meme floating around at the moment where you ask ChatGPT the following and it appears to offer deep insight into your personality: From all of our interactions what …

simonwillison.net

Eduardo Levy Yeyati @eduardoyeyati.bsky.social · Nov 24

Con Luciana Vazquez hablamos de cómo la IA trae oportunidades o riesgos para el país según nos preparemos, por qué Elon & Sam promueven el ingreso universal, y cómo creo que el #hechoxhumanos será la marca identitaria que defina la frontera del trabajo.
youtu.be/MWGj918EO-I?...

Milei y la utopía argentina de la IA: ¿por qué no alcanza con macro ordenada y desregulación total?

YouTube video by LA NACION

youtu.be