Por Lucía Benítez Eyzaguirre

Resumen

La autonomía que van logrando los algoritmos, y en especial la inteligencia artificial, nos obliga a repensar los riesgos de la falta de calidad de los datos, de que en general no estén desagregados y los sesgos y aspectos ocultos de los algoritmos. Las cuestiones de seguridad y éticas están en el centro de las decisiones a adoptar en Europa relacionadas con estos temas. Todo un reto, cuando todavía no hemos logrado ni la soberanía digital.

Abstract

The autonomy that the algorithms are achieving and, especially, artificial intelligence forces us to rethink the risks of lack of quality in data, the fact that in general they are not disaggregated, and the biases and hidden aspects of the algorithms. Security and ethical issues are at the center of the decisions to be taken in Europe related to these issues. It looks like a big challenge, considering that we have not yet achieved even digital sovereignty.

IA y soberanía digital

Los algoritmos organizan y formatean nuestra vida. Como si fueran un software social y cultural, éstos se van adaptando a los comportamientos humanos, y avanzan en su existencia autónoma. Sin embargo, vivimos de forma ajena a su capacidad de control sobre la desigualdad, sobre la vigilancia de nuestras vidas o al margen del desarrollo inminente del internet de las cosas o de la inteligencia artificial (IA): como si pudiéramos darnos el lujo de ignorar cómo se van independizando cada vez más de las decisiones humanas. Ahora, por ejemplo, por primera vez se ha planteado si habrá que modificar los criterios de patentes después de la intención de registrar como propiedad intelectual los inventos y diseños hechos por una inteligencia artificial. De momento, ni la Unión Europea (UE) ni el Reino Unido se han mostrado dispuestos a aceptar una iniciativa de este tipo sin un debate sobre el papel de la IA y del escenario de incertidumbre que esta situación abre.

Es en este contexto que comienza a oírse una pluralidad de voces que piden una regulación de las tecnologías asociadas a la IA; un freno al futuro de un desarrollo autónomo e inseguro. Algunas de las corporaciones de los GAFAM -el grupo que concentra las cinco empresas más grandes en tecnología en el mundo-, como Microsoft o Google ya han pedido esta regulación. Es más, incluso pareciera que estos gigantes tecnológicos comienzan a avanzar hacia la autorregulación en cuestiones éticas o de responsabilidad social, a la vista del impacto que no hacerlo puede tener sobre su reputación. La cuestión para la UE supone valorar y reconocer los riesgos del devenir incontrolable de la IA, sobre todo en asuntos como la salud o la vigilancia. De ahí que parece que el reconocimiento facial en lugares públicos se frenará en los próximos años en algunos países de Occidente, para así prevenir los riesgos detectados en China.

Para combatir los riesgos de la IA hay que comenzar por asegurar la calidad de los datos y los algoritmos, investigar sobre los sesgos que producen y la responsabilidad sobre los errores y criterios. La IA se entrena en muchos casos con datasets no desagregados y a menudo ya sesgados, por lo que conducirá a algoritmos deformados y poco representativos de la población y a desarrollos parciales, de baja calidad y dudosos resultados. Frente al cada vez más numeroso trabajo que se realiza con datos masivos, apenas hay estudios técnicos sobre su impacto humano y social. Por lo mismo, trabajos como los del profesor Matthew Fuller son un clásico recurso para tomar conciencia de la importancia de la transparencia sobre el funcionamiento de los algoritmos. Fuller plantea la aplicación de sistemas que garanticen la condición verdadera de los resultados, la mejora del modelo a partir de un mayor número de conexiones, un funcionamiento que muestre las conexiones sociales o que ponga en evidencia que a menudo se supera la capacidad de los propios sistemas que se analizan con algoritmos.

Si queremos atender a los riesgos de la IA hay que comenzar por el logro de la “gobernabilidad algorítmica”. Este concepto supone la prevención del abuso y del control con el que los algoritmos regulan nuestra vida o con el que la programación rige nuestro quehacer, nuestras rutinas. Esta gobernanza es una garantía de la transparencia, con la supervisión colectiva de usuarios y empresas de los resultados, y la responsabilidad ante el uso de la información. Los algoritmos deben garantizar la transparencia y calidad de los datos (concepto conocido como open data en inglés), ofrecer su propio código de fuente abierto, que sea auditable por sus usuarios y que pueda responder a las reclamaciones fruto de los controles ciudadanos. Pero también es imprescindible que el algoritmo sea leal y justo, es decir, que evite la discriminación que sufren las mujeres, las minorías, o cualquier otro colectivo desfavorecido. Y si se trata de un algoritmo en línea, hay que tener también en cuenta las API (Application Public Programming Interface) públicas porque condicionan tanto la recolecta de datos como la forma en que se aplican técnicas comerciales, que oculta cómo se apropian de la información.

Este espíritu también se recoge en la Declaración de Zaragoza de 2019 a partir del debate de profesionales y académicos sobre los efectos adversos, y los riesgos potenciales. Sin embargo, esta declaración también señala las recomendaciones de uso de la IA, da a conocer sus impactos y su evolución en la sociedad. Esto lo hace a través de cinco puntos sobre las dimensiones humana y social, el enfoque transdisciplinar con el que abordar la AI, la responsabilidad y el respeto a los derechos, a partir de un código deontológico propio.

La Declaración pone el acento en la necesidad de desarrollos para las políticas de interés público y la sostenibilidad, pero siempre a partir de sistemas trazables y auditables, con un compromiso con los usuarios para evaluar el cumplimiento de sus objetivos y separar los defectos o desviaciones. En cuestiones éticas, la Declaración propone la formación de los programadores no sólo técnica sino ética, social y humanista, ya que los desarrollos de software también deben contemplar estas dimensiones, así como diferentes fuentes de conocimiento y experiencia.

La Declaración de Zaragoza también incluye un “derecho a la explicación” sobre las decisiones algorítmicas, siempre y cuando éstas entren en juego con los derechos fundamentales de las personas. A pesar del que el Reglamento General de Protección de Datos de la Unión Europea ha avanzado en derechos digitales, todavía estamos muy lejos de una soberanía tecnológica al estilo de la francesa. Desde 2016, Francia se rige por la “Ley de la república digital” que impulsa los algoritmos auditables, la neutralidad de la red, la apertura de datos, la protección de la privacidad y lealtad de las plataformas con la información de sus consumidores, el derecho a la fibra y a la conexión a Internet, el derecho al olvido, la herencia digital, la obligación de informar de las brechas de seguridad detectadas, las multas en materia de protección de datos.

LUCÍA BENÍTEZ

by Tomás Dodds

Introduction
For over 80 days now, and with no end in sight, Chile has been in the grip of waves of social protests and cultural manifestations with tens of thousands of demonstrators taking to the streets across the country. For many, the upsurge of this social outburst has its roots in a civil society rebelling against an uncaring economic and political elite that has ruled the country since its return to democracy in 1990. Mass protests were soon followed by a muddle of misinformation, both online and in the traditional press. In this blog post, I provide insights into how Chilean activists, including journalists, filmmakers, and demonstrators themselves, have started using citizen-generated data to fight media disinformation and the government’s attempts to conceal cases of human rights violations from the public.

Background
The evening of October 18^th 2019 saw how Chileans started to demand the end of a neoliberal-based economic system, perceived among citizens as the main cause for the social inequalities and political injustices that occurred in the country over the last decades. However, demonstrations were met with brutal police repression and several corroborated cases of human rights violations, including sexual torture. To this day, information gathered by national and international non-governmental organizations show at least that 26 people have died and more than 2.200 have been injured during the rallies.

Although I was raised in Chile, today I am living in Amsterdam. Therefore, I could only follow the news as any other Chilean abroad; online. I placed a screen in my room streaming in a loop the YouTube channels of the prime-time late-night news of major media outlets. During the day, I constantly checked different social media platforms like Facebook or Twitter, and from time to time I would get news and tips from friends and fellow journalists in the field over WhatsApp or Signal. Information started flooding every digital space available: a video posted on social media in the morning would have several different interpretations by that evening, and dissimilar explanations would be offered by experts across the entire media spectrum by night.

And this was only the start. Amidst the growing body of online videos and pictures showing evidence of excessive military force against demonstrators, Chilean President Sebastián Piñera sat in on a televised interview for CNN’s Oppenheimer Presenta where he claimed that many recordings circulating on social platforms like Facebook, Instagram, and Twitter have been either “misrepresenting events, or filmed outside of Chile.” The President effectively argued that many of these videos were clearly “fake news” disseminated by foreign governments seeking to destabilize the country, like those of Venezuela and Cuba. Although Piñera later backed down from his claims, substantial doubts were already planted in Chileans’ minds. How could the public be sure that the videos they were watching on their social networks were indeed real, contemporary, and locally filmed? How could someone prove that the images of soldiers shooting rubber bullets at unarmed civilians were not the result of a Castro-Chavista conspiracy, orchestrated by Venezuelan President Nicolás Maduro, as some tweets and posts seem to claim with a bewildering lack of doubt? How could these stories be corroborated when most of them were absent from the traditional media outlets’ agendas?

As a recent study suggests, unlike their parents or grandparents, the generation that was born in Chile after 1990 is less likely to self-censor their political opinions and show a higher willingness to participate in public discussion. After all, they were born in democracy and do not have the grim memories of the dictatorship in their minds. This is also the generation of activists who, using digital methods, have taking it up to themselves to mount the digital infrastructure that makes relevant information visible and, at the same time, accessible to an eager audience that cannot find on traditional media the horror tales and stories that reflect the ones told by their friends and neighbors. Thus, different digital projects have started to gather and report data collected by a network of independent journalists, non-governmental organizations, and the protestors themselves in order to engage politically with the reality of the events occurring on the streets. Of these new digital projects, here I present only two that stand out in particular, and which I argue help to alleviate, or at least they did for me, the uncertainty of news consumption in times of social unrest.

(Image courtesy of Osvaldo Pereira) 

From singular stories to collective data
Only four days after the beginning of the protests, journalists Miguel Paz and Nicolás Ríos started ChileRegistra.info (or Chile-Records in English), a depository of audio-visual material and information regarding the ongoing protests. Chile-Registra stores and distributes videos that have been previously shared by volunteers and social networks users who have attended the rallies. According to these journalists, traditional media could not show videos of human rights violations shared on social networks because they were unable to verify them, and therefore would only broadcast images of riots and barricades, which would later produce higher levels of mistrust between the demonstrators and the press.

As a response to this problem, the project has two main purposes; First, to create a “super data base” with photos and videos of the protests, and military and police abuses. Second, to identify the creators of videos and photos already posted and shared on social networks, in order to make these users available as news source or witness for both traditional media and the prosecutors. National newspaper La Tercera and Publimetro, among other national and international media outlets, did already use this platform to published or broadcast data collected within the depository. By using this project, users were able to easily discredit Piñera’s claims that many of these videos were being recorded abroad.

The second project I would like to draw attention to is Proyecto AMA (The Audio-visual Memory Archive Project in English). AMA is a collective of journalists, photographers, and filmmakers who have been interviewing victims of human rights violations during the protests. Using the Knight Lab’s StoryMap tools, AMA’s users can also track where and when these violations have taken place, and read the personal stories behind the videos that they most probably saw before online. According to their website, members of this project “feel the urgent need to generate a memory file with the images shared on social networks, and give voice and face to the stories of victims of police, military and civil violence in Chile.”

These two projects have certainly different approaches for how they generate content. While ChileRegistra relies on collecting data from social media and citizen journalists uploading audio-visual material, Proyecto AMA’s members interview and collect testimonies from victims of repression and brutality. Although the physical and technological boundaries of each media platform are still present, these projects complement each other in a cross-media effort that precisely plays with the strengths of each of the platforms used to inform the work activists do.

New sources for informed-activism
These projects are at the intersection between technology and social justice, between the ideation and application of a new digital-oriented, computer assisted reporting. Moreover, the creation and continuous updating of these “bottom-up” data sets detailing serious human rights violations have not only been used to further the social movements, but they also indicate the necessity that digital activist have to gather, organize, classify, and perhaps more importantly, corroborate information in times of social unrest.

As long as Chileans keep taking to the streets, this civil revolution presents the opportunity to observe new ways of activism, including the use of independently-gathered data by non-traditional media and the collection of evidence and testimonies from victims of police and military brutality in the streets, hospitals, and prisons.

What can we, only relying on our remote gaze, learn from looking at the situation going on today in Chile? This movement has shown us how the public engagement of a fear-free generation and the development of a strong digital infrastructure are helping to shape collaborative data-based projects with deep democratic roots.

Lastly, let’s hope that these projects, among others, also shed some light on how social movements can be empowered and engaged by new ways of activism actively creating their own data infrastructure in order to challenge existing power relations, seemingly resistant to fade into history.

TOMÁS DODDS