AI Governance in Argentina and Uruguay

Alex Moltzau
The Startup
Published in
4 min readJun 8, 2019

--

As I was walking on my way back home it struck me that I knew nothing about the developments of artificial intelligence in Latin America. So I decided to commence a surface exploration.

Starting my search I found a report written by the World Wide Web Foundation in October 2018. The report is a study of the implementation by governments in Argentina and Uruguay. So let’s start there.

The World Wide Web Foundation was established in 2009 by web inventor Sir Tim Berners-Lee to advance the open web as a public good and a basic right. We are an independent, international organisation fighting for digital equality — a world where everyone can access the web and use it to improve their lives.

Algorithms and Artificial Intelligence in Argentina and Uruguay

The report focuses on four government cases:

  • A system to predict teenage pregnancy and a system to predict school dropout with the support of Microsoft in Salta, Argentina.
    -Both academic and independent specialists question the methodology.
    -Disclosing information on how the model works would relieve tensions.
    -Publishing protocols and follow-up actions with quarterly protocols would be good.
    -The model has a degree of of explainability, but is only available to public officials.
    -Databases are closed and design closed off to the public which restricts the possibilities for doing audits.
    -It is impossible to know whether the model leads to fair or unfair outcomes.
  • An algorithm to build dashboards outlining commercial opportunities in the city’s neighbourhoods and thus direct investments in Buenos Aires, the capital of Argentina.
    -Simple design with statistical information.
    -There is no data available about its impact and no causal links.
    -There is not enough information on how some outputs are estimated.
    -It would be useful to show how indicators will improve policy.
  • Predpol policing software to predict crimes in Uruguay.
    -There was no information about the error rate.
    -Reduction was detected, but not in overall terms.
    -There is a risk it was used to justify police presence in marginalised areas.
    -The data was not public and PredPol is a proprietary black box.
    -PredPol got replaced by a basic statistical tool.

After going through these four cases the report concludes with three suggestions.

  1. Develop local infrastructure, expertise, and regulation.
  2. Define and adopt practices of transparency, public participation and accountability in the development and implementation stages.
  3. Establish criteria for evaluating the risks of different models and their implementation.

How Can We Question Public Officials Implementing AI?

It also outlines a few questions to public officials in a diagram of the ecosystem in which the algorithms operate:

The proposal focuses on four components:
1. The process of data collection/creation
2. The design of the tools
3. The administrative protocols that surround the tool’s output
4. The legal and social norms that define the broader context in which the policy is executed.

Further the foundation brings home an important point:

…the Web Foundation actively promoted the process of opening government data, because we understand that, in order for a robust democratic debate to take place, people must understand how the public sector works… Ensuring transparency, public participation, and accountability is a matter of urgency. Only this will guarantee that the implementation of new technologies by government institutions will be done in accordance with democratic values and the interests of the people.

This is day six of #500daysofAI.

There are still many grey areas of the field of AI to be explored.

-

What is #500daysofAI?

I am challenging myself to write and think about the topic of artificial intelligence for the next 500 days with the #500daysofAI. It is a challenge I invented to keep myself thinking of this topic and share my thoughts.

This is inspired by the film 500 Days of Summer where the main character tries to figure out where a love affair went sour, and in doing so, rediscovers his true passions in life.

I hope you stick with me for this journey and tell me what you think!

--

--

Alex Moltzau
The Startup

Policy Officer at the European AI Office in the European Commission. This is a personal Blog and not the views of the European Commission.