Jacobo Elosua

UncertAInty

Developing a startup’s strategy has never been an easy job. The world is an uncertain place, after all. And it continuously dishes out myriad crucial variables that, despite all of their efforts, just fall outside any management team’s control.

On top of that, the advent of Artificial Intelligence in the form of advanced –if not always so large– language models, has increased the task’s difficulty by an order of magnitude, I’d say. Forecasting the future has arguably become harder than ever.

Iris.ai provides a fine example. With the company’s origin deeply rooted in the machine processing of scientific text, the unique engineering expertise developed over the years has now positioned the company to launch Neuralith^TM.

Neuralith^TM is an exciting new Retrieval-Augmented Generation-based solution that enables enterprises to unify and leverage all their proprietary data –not just R&D– regardless of its provenance or current state. More a flexible tool for innovative AI builders than a traditional end user-facing software product, it cements Iris.ai’s position in the applied AI B2B space.

In some odd way, this previously unforeseen strategic development evokes the history of telephony in less developed countries –many in Africa–, where poor legacy fixed telecommunications infrastructure was often leapfrogged by the rapid straight deployment of mobile technology. For that is what current AI is achieving: the elimination of redundant intermediate steps at neck-breaking speed!

Check Neuralith^TM out and, if relevant, please come talk to us about it!

 

Luz y calor

Invitado por la infatigable y brillante Mercedes Valcárcel, el pasado miércoles, 26 de marzo, participé en un evento de la EJE&CON, Asociación Española de Ejecutiv@as y Consejér@s, dirigido principalmente a ejecutivas considerando su posible participación en el patronato de fundaciones españolas.

En ‘De la Dirección al Patronato’, Beatriz Sánchez Guitián, Directora General de la Fundación máshumano, con una dilatada trayectoria como patrona a sus espaldas, habló de la importancia de aportar luz y calor. Esto es, buen juicio y apoyo moral a los equipos responsables de materializar el día a día de estas fundaciones.

Además de esa acertada e interesante idea, también conversamos sobre la gestión de la motivación, cómo asegurar el alineamiento entre patronatos y equipos, las distintas vías de acceso a la colaboración, o las cuestiones que un candidato a patrono debe tener en cuenta antes de aceptar cualquier oferta concreta.

Más allá de la gestión de los varios riesgos asociados a formar parte de uno de estos órganos de gobierno, al evaluar lo acertado –o no­– de sumarse a un patronato, también pusimos de relieve que el candidato o la candidata en cuestión debe formarse un punto de vista propio sobre los logros precisos –muchos o pocos– que está alcanzando esa entidad de cara a lograr sus fines fundacionales. Algo que siempre intentamos tener muy presente en Civio.

Se trató de un acto que, a juzgar por las reacciones in situ y posteriores, parece que resultó de interés para el público. Espero que redunde en más y mejores candidatas dando el paso de cara a impulsar la creciente actividad sin ánimo de lucro que lleva a cabo la sociedad civil en nuestro país.

 

Higher standards

Being impressed is one of the best feelings there are.

Sometimes you may read a book that is just too good. Nate Silver’s On the Edge: The Art of Risking Everything and Yuval Noah Harari’s Nexus: A Brief History of Information Networks from the Stone Age to AI are the latest too examples.

Both are ripe with insight. The first one proposing Agency, Plurality and Reciprocity as a forward-looking version of the French revolution’s iconic Liberté, Égalité, Fraternité motto. The second calling out the old established belief that more information equates desirable progress –the ‘naive view of information’–.

In other instances, it might be a blogpost or a podcast. Matt Yglesias’ full of conviction “A Common Sense Democrat manifesto”, Scott Alexander’s just hilarious “Lives of the Rationalist Saints”, This American Life’s refreshing episode on How to Win Friends and Influence People –with the brilliant David Sedaris–, or a particularly pointed interview by Tyler Cowen or Sam Harris come to mind.

And yet, some other time, you decide to write a quick letter to the editorial features editor of The Wall Street Journal –James Taranto himself–, seeking clarification on a point about freedom of speech argued in one of his columns, to find that he engages within the ten-minute mark. Not only that. He does it with remarkable precision and addressing the points raised in a most concise manner.  

Then, slowly, the thought sinks in. All of them have one thing in common: high standards. An entire culture that underpins that demand –inner and outer– for top quality output. And that makes me envious, but also motivates me to want to do better every single day.   

 

RSpace™ 1.3

Late last year the Iris.ai team announced the release of the latest version of our platform, RSpace™ 1.3.

The new product functionalities include a monitoring feature, new integrations –incl. OpenAlex–, the ability to chat with full text articles, the auto-syncing of datasets, versioning, or support for new file formats –incl. .jsonl and .doc–.

You can read more about the announcement here.

These news have been quickly followed by today’s unveiling of the chat tool’s latest upgrade: the R&D assistant system is now capable of handling ten references simultaneously.

Combined, these enhancements should significantly boost users’ research productivity and collaboration.

I recommend you give it a try!

 

España y la AI

El economista Jesús Fernández-Villaverde es una de las mentes más brillantes de España. En un desierto de perezosos dizque análisis reciclados y a remolque de otros, sus artículos y conferencias suponen un oasis de pensamiento sólido, valiente, estructurado y personalísimo. Dicho esto, su tono soberbio y en ocasiones faltón creo que puede minar innecesariamente el alcance de sus mensajes.

Este mes de diciembre, con motivo de la conferencia anual que da desde hace años en la Fundación Rafael del Pino, Fernández-Villaverde publicó un artículo –conjuntamente con Luis Garicano y Tano Santos– en el que abogaba por una agenda de crecimiento económico. Es decir, por un aumento radical de nuestra productividad, impulsado por la adopción y generación de tecnologías punteras

En concreto, y tras lamentar la pérdida de oportunidad que supuso la revolución industrial del siglo XIX para España, Fernández-Villaverde y sus coautores fijaron su vista en los desarrollos de la inteligencia artificial y el masivo potencial de transformación que éstos ya habilitan y seguirán impulsando.

Plenamente de acuerdo con el mensaje expresado, me llamó la atención que los autores hicieran un juego correlativo con el libro de historiografía El fracaso de la revolución industrial en España, 1814-1913, cerrando su artículo con la frase: “No debemos esperar a que un historiador futuro escriba: El fracaso de la revolución de la inteligencia artificial en España, 2012-2111.”

¿Por qué? Si cien años parece un intervalo adecuado para analizar y atribuir el malgaste de la antigua ventana de oportunidad, prácticamente podemos asegurar que la actual permanecerá abierta un periodo de tiempo mucho menor.

La aceleración de los desarrollos tecnológico que están facilitando la explosión de la inteligencia artificial apuntan a que los ganadores y perdedores de esta revolución se podrán declarar mucho antes del año 2111. De hecho, en algunos mercados –p.e. Metaculus– la predicción mediana sitúa el alcance de la inteligencia artificial general en 2031.

¿Y cuáles son las señales de que España esté afrontando este reto clave con alguna garantía de éxito? Sinceramente, creo que ninguna.

Mientras los investigadores nacionales continúan a la espera de pequeñas ayudas a la investigación, con sus interminables retrasos, OpenAI acaba de lanzae su modelo o3. En el benchmark principal de matemáticas –el ARC-AGI, dificilísimo incluso para matemáticos profesionales– al principio del año los modelos de inteligencia artificial obtenían una nota del 5%. En septiembre, del 32%. Desde este mes de diciembre, el listón está ya en un 88%.

Así se escribe la historia. Unos la protagonizan; otros, por desgracia, siguen mirando el dedo.

 

Paliar los efectos de la DANA

Hoy se cumple justo un mes desde que una DANA de fuerza inusual causó estragos en Valencia. Como en ocasiones anteriores –p.e. la pandemia del Covid-19–, el equipo de Civio se puse manos a la obra inmediatamente después.

¿El resultado? Una guía completa –otra– y continuamente actualizada –ojo, muy importante– que recoge todas las ayudas disponibles para las personas afectadas. Puedes ayudarnos a darle difusión compartiendo estos enlaces en Instagram o LinkedIn, por ejemplo.  

Y, si te motiva el trabajo de la fundación, por favor considera convertirte en socio. ¡Te necesitamos!

 

We will miss you dearly, Helen

On Friday, October 18^th, our friend Helen Darbishire left us. Founder of Access Info Europe and at the center of an endless list of civic initiatives –including Coalición Pro Acceso here in Spain–, she was a defining figure in advancing transparency. Both in our country and internationally. 

Helen was one of a kind. Smart, hard working, inspired, generous collaborating, always good spirited... Before leaving us so prematurely, she was the best embodiment I have come across of a true activist. Those who understand the sheer magnitude of the challenge faced and yet commit themselves relentlessly to it — to making a difference. 

Here is a heartfelt good bye text from the team at Civio –in Spanish–. She was instrumental in getting the project launched and well on its way over a decade ago. Helen will be dearly missed by so many, with me and Civio's entire team counting ourselves definitely among them.  

Family and friends will be honoring her life –through a get together in Madrid– on Saturday, November 16^th. 

 

 

Leveraging deep knowledge

When we started Iris.ai back in 2015, our vision was to make science more accessible and actionable for everyone. Starting –you always have to start somewhere– with professional R&D personnel.  

Much has changed since then. Large Language Models –LLMs– are now at the center of an ironic development: the better they become, the higher the bar is raised when it comes to assessing their intelligence.

OpenAI’s new model –o1– has been tested at an IQ of 120. Higher than 91% of humanity. And the pace of algorithmic improvement is not showing any signs of slowing down. Around such developments, a new layered ecosystem of AI players is beginning to emerge.  

In this context, as we enter into a new phase of the company’s development with a freshly raised funding round, the time has come to scale Iris.ai’s capabilities. To go beyond building narrow scientist-focused research applications. Applying all of our engineering prowess, infrastructure, and tools towards the transformation of unstructured deep knowledge

We want to radically change how companies work with expert documentation, external and internal. Transforming millions of complex pieces of information –aka deep knowledge– into much more actionable innovation.

Here you can have a peek at our new website:

Does salty beat decent?

The reactions to Kamala Harris’ first TV interview as the Democratic presidential nominee have been quite enrichingly diverse.

This one in the New York Times caught my eye. It was more positive that some of the others published by the same newspaper, which connected with the point made recently by Nate Silver, during the media tour presenting his newest book. Namely, that the Old Grey Lady systematically speaks to its centrist and progressive-leaning audiences at the same time.

In the opinion piece, Michelle Cottle argued that Harris’ saltiness contrasted not only with the utter indecency of Donald Trump, but also –yes– with President Biden’s –by now legendary– decency. This observation was framed as a net positive remark.

When did being earthy and crude surpass exhibiting consistent standards of respectable moral behavior as a leader, I was left asking. Next week I turn fifty. For a second there I felt much older, though.

Civil society’s ailments

A weak, someone-else-will-do-it, civic culture. Too few resources –not-for-profit grants and donations– at its disposal. A certain lack of professionalism and adherence to strict standards, engaging in frowned upon practices when done by others. That is the triad of factors –root causes–, which, in my view, constrains the advancement of a healthy civil society in Spain.

It is also the reason why I feel so proud of David Cabo and Eva Belmonte’s amazing work leading the Civio team. Because it places all the emphasis on educating the public, treating readers like grown up adults. Because it has always been sustainable whilst continuing to grow its footprint. Because it never takes shortcuts or ducks hard choices. Even when it implies being unpopular — for that is the real price of being fiercely independent.   

Please check out Civio’s latest campaign and consider becoming a member!