Enthusiasm and complexity: Learning from the “Prometea” pilot in Colombia's judicial system
At the start of 2019, the arrival of a so-called artificial intelligence (AI) solution for the Colombian Constitutional Court was announced in the media. Many questions were raised by civil society and academia regarding the likely impact of “Prometea” – the name of the system – including how it worked and the decision-making process that led to its adoption. There was, however, little information forthcoming. Prometea ended up being a pilot that is currently on hold.
The arrival of Prometea presented us with two possible levels of analysis: 1) the impact of the AI system itself, and 2) the impact of the adoption of an AI system in a society. Both levels are at risk of being overlooked due to the “charismatic halo” surrounding technologies and the false perception of technological neutrality and its infinite potential to solve pressing problems.
By considering the case of Prometea, we aim to identify some of the possible social and human rights impacts that are the result of the adoption of AI-based solutions in particular contexts. For this purpose, we reviewed how the media covered Prometea and conducted a series of interviews with different actors involved and interested in the issue.
In 2011, the Colombian Congress enacted the Code of Administrative Litigation and Procedures. This code required (in article 186) the Supreme Judiciary Council to digitise its records within five years. This time has now expired with no significant advances in this regard, despite the fact that several other laws, some going back as far as 1996, have also required the creation of digital judicial records.
Previous to the creation of the Code of Administrative Litigation and Procedures, the Colombian judicial system had been troubled by delays in resolving cases in part due to legal processes being paper-based. Among the risks of paper-based records were the possibility of losing the records, the vulnerability of the records to being destroyed, for example by fire, and their deterioration over time. Processing cases on paper also affected the rights of plaintiffs because it took time to process the large numbers of paper-based records needed in each legal action.
According to statistics published by the Private Council for Competition in 2018, the Colombian judicial system commonly takes between 385 to 956 days to settle a case, and depending on the issue at stake it may take even twice as long as that. One of the main recommendations the Council gave to the national government was to digitise judicial records as a means of achieving not only efficiency in justice, but also timeliness.
In December 2018, it was announced by the national press that an agreement had been reached by the highest national courts and government to digitise judicial records. This agreement was nothing different from the previous year’s promises and provisions in the law on the same issue that remained unattended. But an announcement made in the press during February 2019, about an AI solution that would be applied to the national judicial system – specifically in the Colombian Constitutional Court – caught everyone’s attention.
Prometea was presented as the main means of resolving the inefficiency and delays within the Constitutional Court. Its design, according to the media coverage, would impact the selection and revision process involving the writ of protection of constitutional rights that every citizen has the right to file to protect his or her own fundamental rights without any cumbersome legal formality.
The court receives more than 2,000 writs of protection daily coming from all the judiciary benches across the country. The Constitutional Court only has nine judges and under 200 employees who serve the entire country. In the selection process, the court decides which cases, due to their relevance and novelty, may need to be reviewed by one of these judges to decide whether or not to protect the allegedly violated fundamental rights.
Media coverage and social impact of Prometea
The media reported on Prometea in early 2019 when the Constitutional Court announced the conclusion of its pilot to help assistants and court personnel sort, read and retrieve key information from the hundreds of cases received to be reviewed. The coverage of Prometea was scarce, but had a significant impact.
There was a radio interview with the president of the Constitutional Court, a very short segment about the system in a well-established TV news slot, a piece in a major newspaper, and two articles in a specialised magazine covering judicial and legal issues. The most nuanced and critical pieces about Prometea appeared in the specialised magazine. They considered the needs of the court and alternatives to the proposed system, such as the redesign of the selection process, and noted that there was not much information available publicly about Prometea. In contrast, the rest of the media was uncritical and plain when covering the pilot of Prometea.
Several points can be made about the media coverage of Prometea. First, the view that the selection process needed to be improved was shared by many actors and was in most cases the angle of the news stories about the system. However, there was no explanation about what Prometea was, what it does and how it does it. Second, in every instance it was presented as an “artificial intelligence” solution, the operational readiness of the system was highlighted, and the fact that it was employed in a pilot that only focused on the right to health was hardly mentioned. “With Prometea, the Constitutional Court finally enters the world of the highest informatics technology,” said one TV report on the pilot.
The charismatic effect of AI systems was felt in the comments on Prometea by some law scholars. Grenfieth Sierra Cadena, head of public law research at the Universidad del Rosario, emphasised the importance of the project because it was the first time AI had been applied “in an executive and active way by a supreme court.” In the same piece, Cadena stated that “the court improved the management of writs of protection of constitutional rights related to the right to health by 900%,” a number that apparently is calculated using the time that it takes the system to create documents, but does not include time saved in the process of selection. This is striking, as several parts of the software (i.e. the search functions and the automatic document generator) seem to be confused and mixed instead of clearly discerned for the purposes of evaluating its efficiency.
The media coverage showed two key characteristics: limited information and excessive enthusiasm. The news was clear enough: the Constitutional Court had adopted an AI-based solution to overcome the bottlenecks in the selection process. However, the lack of any further information or investigation by the media seems to be for three possible reasons: 1) a blind trust in information coming from the Constitutional Court, 2) concerns that looking deeper into the issue may be read as mistrust of the Court, and 3) the complexity of the issue, both for the journalist and for the reader.
This situation had a double-edged effect – muting the possibility of debate, and at the same time, stimulating the need for more information. While any possibility of public debate among the general population was defused, concerns were raised by groups interested in AI, mainly in civil society and universities. This increased the demand for further information and for opening a public debate about Prometea. The few events that took place in universities to try mitigate these demands only increased them. Cadena, acting as the promoter of Prometea at these events, towed the same line taken by the media, and did not provide more technical information on the system.
This led to social and human rights concerns on three levels: 1) general concerns regarding AI, 2) concerns regarding the relationship between digital technology-based solutions and the Colombian context, and 3) and concerns regarding the selection process at the Constitutional Court. These concerns are related to transparency, privacy and data protection, and include issues to do with labour rights among other secondary concerns.
Prometea impacts transparency in two ways: first, regarding how a technological developer is selected in terms of the framework for public contracting, and second, how stakeholders, scholars and civil society organisations are allowed to participate in the process of discussing and deciding on technological solutions intended for the judicial system. As pointed out by scholars, a key factor in this is understanding exactly what the proposed technological solution is, and how it works.
It is also important to know the legal rationale behind the decision made by the Constitutional Court to implement the system, and how use of Prometea can be reconciled with normal protocols for motivating changes in the judicial system. The lack of transparency with regards to the AI decision-making process and how it may affect a citizen's right to due process is also worrisome. Furthermore, due process allows the writ of protection plaintiff to file an appeal in case he or she is not favoured in the Constitutional Court decision to review their case. A technological solution may put in danger the exercise of this right because it would be impossible for the plaintiff to argue against a machine’s decision irrespective of how the algorithm and system work.
In terms of privacy and data protection, the main concern has to do with sensitive data being shared with third parties, such as a software developer. The fact that minors are involved in some cases, or others are to do with sexual crimes, among other situations that may require the anonymity of victims and their personal information or data, is considered critical. It is a breach of confidentiality for someone other than the judge and the parties involved in the processing of the case to access this information or data. It is especially worrying that a possible leak of personal data to the media or other third parties with an interest in the case can occur given the system's vulnerability in this respect, with irreversible consequences in terms of the protection of privacy for those involved in the case.
On labour rights, Prometea was seen as a replacement for those doing basic clerical activities, but by no means affecting more specialised work such as that done by judges or even judges' assistants. At the Constitutional Court this impacts on the so-called ad honorem system, training and experience for law students who help with writ of protection selection: reading files, writing abstracts, and classifying cases. This may also have an indirect impact on the future staffing at the court, as ad honorem is not only a training ground for future constitutionalists, but also a first step into more important positions at the court, such as judge's assistants.
After researching Prometea, we believe there are three areas of consideration when adopting AI-based solutions. First, the enthusiasm in adopting these systems overshadows the need to evaluate the human factors involved, both with regard to the skills needed to use the technologies, and issues to do with a reluctance to change. Second, the limitations of the technical infrastructure need to be considered; and third, the flexibility of processes from a legal point of view is important – or how they can be adapted properly to a technological platform.
With respect to how we adopt a technological solution in the Colombian judicial system, Prometea showed that the discussion is led by lawyers with few to no computer scientists or technical people involved. This inevitably results in a shallowness in the discussions due to the complexity of AI systems. The decision-making process focused on big and vaguely defined problems with the aim of solving these with an “AI solution”.
Such concerns lead to questions that are still without clear answers. For instance, should the information which feeds an AI system be included in the concept of “data” according to the Colombian data protection law? Is it time to reassess or “update” our legal vocabulary and frameworks related to digital technologies? Should we think about other, more primary technological solutions such as the digitisation of judicial records instead of more questionable solutions such as AI systems?
As an alternative, some initiatives in Colombia like the Legal Design Lab at the Universidad de Los Andes advocate for a more detailed definition of problems in order to address them with tailored actions.
Finally, when it comes to an AI solution being applied to a specific issue, the usual concerns relating to the use of technologies (the possibility of bias, the opaqueness of the technology, privacy concerns regarding data sets, etc.) were also raised in the case of Prometea. Due to the scarcity of information on Prometea, these concerns were largely speculative, given that it was not possible to understand if these issues were in fact present in this particular case.
Bearing this in mind, we believe that there are several factors that need to be taken into consideration when adopting AI-based solutions:
- Lawyers should be more involved in technical discussions and should advocate for multidisciplinary spaces for discussing the proposed technology.
- A “big problems – big solutions” enthusiasm in the approach should be replaced with a more grounded methodology based on detailed and complex definitions of problems and smaller, tailored proposals. Such methodologies should include an analysis of alternatives: AI systems are just one possibility among others.
- General contextual assessments such as a baseline study should be done, including the capacity of the staff to appropriate the technology, the actual capabilities of the technology, the flexibility of the legal framework to accommodate the technology, the need for technological training for legal practitioners, and the general technological culture and awareness of technology in society.
- More information is needed both for the general public and third parties such as other judicial systems in Latin America, including the Inter-American Commission on Human Rights, where Prometea is being promoted. More information is also needed to inform public policy decision-making processes that affect the administration of justice at the national level.
The following advocacy steps are suggested for Colombia:
- Build interdisciplinary networks ideally involving universities and civil society organisations to enrich the debates around AI initiatives in Colombia.
- Promote transparency in the form of participatory processes as a prerequisite for any AI public initiative.
- Advocate for a detailed and down-to-earth definition of a problem to counter enthusiasm for ready-made AI solutions and to contribute to informed public debate on an issue.
- Request the presence of computer scientists, data scientists or similar experts in conjunction with scholars from human sciences working on technology issues in debates involving the use of AI in public initiatives.
- Fight the preconceptions of technological neutrality and technological determinism that prevent the critical analysis of any solution based on digital technologies.
 In particular, we have interviewed journalists, scholars and constitutional court workers. Given the size of the piece, we have decided to agglutinate the opinions on the main issues raised instead of directly quoting individuals.
 Código de Procedimiento Administrativo y de lo Contencioso Administrativo, Ley 1347 de 2011. www.secretariasenado.gov.co/senado/basedoc/ley_1437_2011.html
 Ley 270 de 1996, Ley 1437 de 2011, art. 186, Ley 1564 de 2012, Ley 1709 de 2014, Ley 1743 de 2014, Decreto 272 de 2015, Decreto 1069 de 2015, Decreto 1482 de 2018.
 Consejo Privado de Competitividad. (2017). Informe Nacional de Competitividad 2017-2018. https://compite.com.co/informe/informe-nacional-de-competitividad-2017-2018/
 Ámbito Jurídico. (2018, 13 December). En estos cinco procesos se implementará el expediente electrónico. https://www.ambitojuridico.com/noticias/tecnologia/procesal-y-disciplinario/en-estos-cinco-procesos-se-implementara-el-expediente; Redacción Tecnósfera (2018, 12 December). Arranca piloto para implementar el Expediente Digital Electrónico. El Tiempo. https://www.eltiempo.com/tecnosfera/novedades-tecnologia/gobierno-anuncia-plan-piloto-de-expediente-electronico-digital-304654; Redacción Judicial. (2018, 28 November). Consejo Superior de la Judicatura aboga por una justicia digital. El Espectador. https://www.elespectador.com/noticias/judicial/consejo-superior-de-la-judicatura-aboga-por-una-justicia-digital-articulo-826076
 Redacción Judicial. (2019, 5 February). Prometea, la nueva tecnología para la selección de tutelas en la Corte Constitucional. El Espectador. https://www.elespectador.com/noticias/judicial/prometea-la-nueva-tecnologia-para-seleccion-de-tutelas-en-la-corte-constitucional-articulo-838034
 The Spanish legal term used in Colombia is acción de tutela. It refers to the right to ask a judge to protect a fundamental right from an imminent threat when no other legal recourses are available. It is similar to what is known in other Latin American countries as a writ of amparo. See: https://en.wikipedia.org/wiki/Recurso_de_amparo
 https://twitter.com/CConstitucional/status/1141011549387153408; https://www.youtube.com/watch?v=r1ifDdWuW-k&feature=youtu.be; Corte Constitucional. (2019, 17 June). La Corte Constitucional está al día en la radicación de tutelas. www.corteconstitucional.gov.co/noticia.php?La-Corte-Constitucional-esta-al-dia-en-la-radicacion-de-tutelas.-8741
 Redacción Judicial. (2019, 5 February). Op. cit.
 Giraldo Gómez, J. (2019, 12 April). Prometea: ¿debe rediseñarse el proceso de selección de tutelas en la Corte Constitucional? Ámbito Jurídico. https://www.ambitojuridico.com/noticias/informe/constitucional-y-derechos-humanos/prometea-debe-redisenarse-el-proceso-de; Rivadaneira, J. C. (2019, 22 March). Prometea, inteligencia artificial para la revisión de tutelas en la Corte Constitucional. Ámbito Jurídico. https://www.ambitojuridico.com/noticias/informe/constitucional-y-derechos-humanos/prometea-inteligencia-artificial-para-la
 Giraldo Gómez, J. (2019, 12 April). Op. cit.
 Rivadaneira, J. C. (2019, 22 March). Op.cit.
 On March 12, a public event was held at Los Andes University to critically debate about the pilot due to the concerns of some professors there. The general characteristics of Prometea were presented (it was an AI solution, it included blockchain technology) but all questions, coming from both the public and the participants on the roundtable, regarding technical details of the pilot (dataset structure, type of algorithm, possible bias in the design or criteria selection, etc.) were eluded. A report on the event is available online in Spanish at: https://gecti.uniandes.edu.co/images/pdf/PROMETEA_EVENTO.pdf
This report was originally published as part of a larger compilation: “Global Information Society Watch 2019: Artificial intelligence: Human rights, social justice and development"
Creative Commons Attribution 4.0 International (CC BY 4.0) - Some rights reserved.
APC Serial: APC-201910-CIPP-R-EN-P-301
ISBN APC Serial: APC-201910-CIPP-R-EN-DIGITAL-302