How can we build child-friendly AI in Latin America and the Caribbean?

Reflections from participants of the AI and Children Workshop in São Paulo, Brazil

The Global Insight team
31 March 2020
This work is part of UNICEF's AI for Children project
Expert Q&A  |  8 minute read
These interviews were conducted during an AI and Children regional consultation that took place in São Paulo, Brazil in March 2020 — the fourth in a series of regional workshops designed to develop policy guidance for AI that protects child rights. Transcripts have been edited for length and clarity.

Photo of Veridiana Alimonti


Veridiana Alimonti
Latin American Senior Policy Analyst,
Electronic Frontier Foundation



What are the challenges to more child-friendly AI in Latin America and the Caribbean (LAC)?

Veridiana Alimonti: Children’s development and empowerment are key elements for conceiving AI systems in line with human rights, although it is not always directly part of the conversation. Bringing this perspective is an important challenge UNICEF is taking up and entails building bridges between general and particular approaches regarding AI strengths and risks. Protection before unfair and deceptive practices, data collection and processing abiding by proper standards and rules, sound processes and means to spot and tackle bias and discrimination — all are general challenges that can have a specific and particularly detrimental impact on children’s lives. 

What do we need to do so that AI empowers children in this region?

VA: Children, parents/family, educators and other related relevant parties form the broader ecosystem through which answers and roles can be developed, combining protection and agency. Issues like AI-empowered surveillance systems and automated decision-making also raise concerns regarding children’s well-being and life opportunities. Increasing surveillance in schools, for example, can actually decrease students’ perceptions of safety, equity and support or disproportionately impact vulnerable groups. And properly addressing transparency, due process and accountability in algorithmic decisions that affect lives is both a major challenge and critical factor if it is to have AI systems empowering children and children’s rights in policy-making in LAC.


Photo of Yves Bastos


Yves Bastos
Head of International Expansion,



What are the challenges for greater child-friendly AI in Latin America and the Caribbean, or specifically in Brazil?

Yves Bastos: I believe we are still failing to bring children not only into AI, but the technological world in general. We still struggle in more basic aspects, such as universal access to basic and quality education and ensuring equality of opportunity so these children can have the chance to develop in a healthy way. For this reason, it’s difficult to skip this discussion and go directly to AI for children because there’s a larger discussion before that waiting to take place. 

As long as we don’t have strong pillars such as quality and inclusive education that bring computing and AI to the basic curriculum, and as long as we don’t have stability and economic inclusion, kids will sadly be on the sidelines with their rights diminished, because they are not being a valued part of the society and are therefore ignored by this market. It’s a reality that we need to urgently change because it’s almost impossible to break this cycle without investing heavily in education and science so we develop a thriving and healthy ecosystem.

What do we need to do so that AI empowers children in Latin America and the Caribbean, or specifically in Brazil?

YB: AI is a complex field, both in technological terms and in how we leverage that capacity to actually achieve palpable, impactful and positive change that could not be done otherwise. Impact by itself is not enough as we have a large number of examples that have done so in a very negative way, worsening people’s lives instead of the other way around. We need to switch the logic: not coming from a technological point of view (e.g., “I have this technology, what can I do with it?”) but from a pragmatic way based on the big challenges we have right now, and they are neither small nor scarce. (e.g.,“We have X reality that we know needs to change and we have these kids from this community whose best interest are a priority for us. In what way can I work with specialists and policymakers to improve these lives and, in this journey, what small part of that can be made better by using AI?”) 

Kids’ development is a fiercely complex issue with no single answer — AI has a huge part in empowering children as a means but will never be the only answer or the end in that mission.


Photo of Cordel Green


Cordel Green
Executive Director,
Broadcasting Commission of Jamaica



What is missing from conversations on AI and children?

Cordel Green: I want to emphasize the capacity of children to maneuver responsibly and productively in the digital economy and society. There is a huge mistake that is being made — children are born wired for the age, and adapt very easily to what is the existing or emerging environment. But what is different this time around is that there is a lot of default action that is being taken, which is potentially as dangerous as it is empowering, so there is a need for some balance. 

So given the liberating potential of technology and the opportunities for the transformation of society, the disruption of models that invite new thinking, it is right that children should be allowed to bask, grow, blossom in that environment. But we should not make the mistake of assuming that because they bask in the environment, adult responsibility goes out the window. You see it in simpler examples such as parents who say to their children, “Come and set the password on my device.” They don't tell children their bank account numbers lightly or their social security numbers,but they share anything that has to do with technology.

We're not paying attention until someone comes along and tells us that there is a problem. That is a huge problem because children need parental guidance in this environment. And as their acquaintance with smart machines and AI gets closer, that input from parents becomes even more important because there is a risk of dehumanizing children. It is critical that adult digital literacy complements digital literacy for children, because you do not want to  stifle the ability of children to innovate and benefit from technology, which is as empowering as it is harmful.

What do we need to do so that AI empowers children in Latin America and the Caribbean?

CG: We need to become more literate about artificial intelligence and its potential and the reasonable demands that we should make to ensure that we have appropriate governance frameworks in place that address ethical issues, such as accountability to ensure that there is transparency. We ought to know who are the people who are infusing values in AI systems and impose on those persons obligations to privilege the rights of children. This covers a whole range of things: not manipulating children, not creating AI solutions that are intentionally addictive, not commercializing their data, not being unconcerned about the appropriateness of the AI solutions we create for them, paying attention to things such as intensity ratings for these solutions — how intense will the experience of children be when they engage with intelligent machines and artificial intelligence?


Photo of Armando Guio


Armando Guio
Affiliate, Berkman Klein Center for Internet & Society,
Harvard University



What topics or issues are missing from conversations on AI and children?

Armando Guio: I think it is very important to look at children as relevant actors on different discussions regarding AI systems implementation. Discussions over the economic impact of technology or the future of jobs should also consider children's opinion and interests. We should not consider this as complex conversations that do not accept their opinions or judgements. It is our task and challenge to make them part of the conversation and provide an adequate weight to their opinions.    

What are some recommendations to empower children as they interact with and utilize AI?

AG: We can empower children by defining a set of interests they share regarding the design and implementation of AI systems. Therefore, the task should be to define a methodology and criteria to identify these interests and not to impose on children those interests and opinions that we believe they have. Likewise, I think that interacting with children from different countries is a very valuable exercise for them to have a more global perspective of these topics. We need to set spaces for global discussions among children to take place. This kind of dialogue can empower them as a generation.


Photo of Patricia Peña


Patricia Peña
Fundacion Datos Protegidos 


What do you see as the challenges to having more child-friendly AI in Latin America and the Caribbean?

Patricia Peña: Artificial intelligence as a new model of society sometimes seems inevitable. Many times, the regulations are very late because the industry has been very quick in the development of solutions or innovations. Academics and NGOs — civil society — act as mediators between sectors that probably don't speak or don't debate enough. In Chile, we're in a very strategic moment right now because the government opened the discussion about what the country needs and wants for AI public policy. It is important to have principles at the center during this moment. Then we have opportunities to make this process more inclusive — this is a huge opportunity to include children. Move beyond the idea that we adults know what is going on (and this typical approach of child protection) and embrace the fact that children are indeed users, already living with this technology. 

What do we need to do so that AI can empower children in the region?

PP: Artificial intelligence for what or for whom? Designed by whom? For every country it is an opportunity to address economic, social, educational, economic problems or gaps. It's a crucial moment, because 10 years ago we were asking the same thing about the role of the internet. It came as a tool — as the promise of development, of access to knowledge, access to education, etc. — and yet we are still dealing with digital gaps in our region. 

We are particularly understanding rights like privacy and personal data. AI is collecting data, processing data from us, from children. So when we talk about how we guarantee privacy for children, their whole life is already online, it’s digital. The footprint, the cycle of that digital footprint, how is it going to impact them? How are we dealing with the impacts? The consequences of big data platforms that are processing their lives, that are processing decisions that are going to affect them, in aspects from education, health — even their sexual orientation.

Latin America doesn't have a strong institutionality of protecting personal data. Today we are facing a generation that not only interacts with video games, applications or a system that constantly monitors how they connect, what they ‘like,’ what digital footprints they leave, but they also need to understand that when they interact with that platform, with that screen, with that system, it is extracting things from them.