Regulations for public procurement of AI systems and new technology for managing and automating public procurement are unfolding together, guided by the need to make public services more efficient, competitive, transparent and accountable.
Regulations for public procurement of AI systems and new technology for managing and automating public procurement are unfolding together, guided by the need to make public services more efficient, competitive, transparent and accountable. New so-called "e-procurement1" systems are being developed, deployed and regulated to solve problems like fraud, lack of transparency and control, and the delay in developing e-government tools and practices in developing countries. In this context, there is an opportunity to reimagine public sector AI, especially when e-procurement systems can be designed to level up the requirements for suppliers by design, after regulation towards a fairer supplier's harvest.
Improving public sector procurement and the automated systems being used could include increasing participation of women and minorities in the suppliers' list, or presenting diversified datasets prior to the bidding, or limiting the involvement of competitors with inadequate focus on human rights, and all of these can be wired and automated into the development of e-procurement platforms. This offers an opportunity to improve public sector AI deployments and refine procurement processes overall. E-procurement systems can be leveraged to enhance results in gender policies, but it will demand a shift in the ethos of implementing e-procurement systems: it is about moving beyond the avoidance of fraud, improving the pace of technology adoption in the public services, and increasing transparency and accountability. Advocates working for better procurement processes should think about feminist approaches, embedded directly into "emergent technologies" frameworks and solutions and some already have already begun to think about how to improve data collection with the aim of supporting more inclusive procurement processes2.
Additionally, there are several elements for developing systems that can aid in improving procurement for public service AI, and three essential recommendations for building equality from scratch when designing e-procurement systems: civic participation, automation of reparation rules, and the constant improvement of the e-procurement platforms:
Brazil recently developed regulation for "electronic" public procurement. The law known as "8666"from 1993 defines rules for bidding and contracts of public administration but abstains from expressing specifications for purchased technology. In 2002, the electronic floor regulation described technology suppliers and some conditions to be met by the latter to ensure better technology services at better prices. Although such regulation introduced some transparency to procurement in Brazil, there is a mismatch between the developments of AI for public service and what has been acquired, which privileges technology that does not leverage equality among genders. The lack of grievance systems and nor a clear channel for civic participation solidifies an approach that overlooks diversity in public service AI. The lack of civic involvement in any public procurement platforms does not incentivize inclusion.
Automation of reparation rules
Governments can implement temporary special measures, creating mechanisms similar to reparation rules implemented in different contexts for correcting historical errors. E-procurement systems can be designed to compensate for the historical lack of gender balance imprinted in public procurement for technology in general, amending historical injustices towards gender equality and mitigating public biased systems in the future. As it is true in different situations where public procurement divested funds from biased marketplaces, this strategy can reshape AI for public service creating new, innovative services that can benefit marginalized communities in the long run.
Examples could be drawn from other areas. In a case from the 2000s,3 a Brazilian city hall adopted rules for buying organic food from cooperatives in the Landless Rural Workers' Movement (MST), encouraging farmers who produced organic food when acquiring supplies for the public school meals that are provided for free in the country. A decade later as a result of these incentives, cooperatives are known to be the biggest producers of organic rice in Latin America, established with the public incentives to create new opportunities for new markets. Similarly, the allocation of economic incentives towards women can create new opportunities for minorities working with AI for public service, prioritizing demands from women and other minorities in communities at the margins.
Constant improvement of the e-procurement platforms
When technology is implemented in a public process, it should be considered part of a cyclic process that should embrace public participation as an input to improve the platform itself. Automated decision-making systems designed to enhance public service AI through better procurement should transcend the monolithic vision of software development as a product and be considered part of a democratic process.
It is essential that a) the software used to develop public procurement systems is open-sourced; b) changes are implemented per citizen request, with issues and new features well documented in public repositories, decided in the communities affected by the public services; and, lastly c) community advocacy is taken into account, leveraging public participation in the development of e-procurement systems.
Actively embedding feminist principles in technology systems for governance is a practice that can fast-track gender equality, and in turn impact thousands of people in the long run. Transparency, agility, and efficiency can be achieved side-by-side with gender equality if the development and implementation of automated decision making systems leverage feminist principles by design. In the same way, e-procurement is deployed to achieve better public service AI can create economic incentives that promote gender equality, consequently giving way to AI systems that prioritize feminist principles.
For governments and public service, the so-called fourth industrial revolution means replacing analogic systems with digital technology for the benefit of the citizens. Governmental software will serve as the spine of democratic procedures and therefore be present on many platforms deployed in developing countries. Public procurement will evolve into a large ecosystem interacting with people beyond companies. This is an opportunity to use emerging technologies like AI, combined with intelligent policies, to support gender equality creating a cascading effect that results in benefits not yet explored by e-government innovators.
From f<a+i>r network discussions:
Feminist methodologies for improving AI procurement