This white paper recommends the inclusion of data ethics requirements in the public procurement of AI-based services and solutions. To this end, a procurement model should include a due diligence process. This makes it possible for the public procurer to gain qualified insight into the planned AI model, its risks and impacts. It would also provide the public procurer with important information about potential suppliers in the market, their skills, methodologies and management systems. Due diligence helps ensure that all relevant risks and impacts relating to data ethics are identified and handled, and that the development, deployment and maintenance of AI accommodates the requirements of a trustworthy AI-based system or solution. As such, a properly diligent procurement process spans the whole lifecycle of the AI-based system or solution.
Recommendations are made in relation to five phases of the procurement process:
- Preliminary risk and impact assessments
- Preliminary screening of potential suppliers
- Contract performance conditions
- Contract implementation.
All recommendations address how to include data ethics in the development and operations of AI-based services and solutions, identifying the relevant processes, tools and methodologies to integrate data ethics requirements and demonstrate their fulfilment.
The need for a common standard
This whitepaper demonstrates the need for a common standard for data ethics within the EU for the public procurement of AI-based services and solutions. Such a step would further a human-centric, sustainable usage of AI in the public sector and provide leverage to a business community that is eager to sell AI-based services and solutions but lacks insight into fundamental rights, IT security and data ethics.
This whitepaper suggests that the following initiatives are integrated and promoted within the European Union’s Public Procurement Strategy.
These recommendations encompass the integration of data ethics principles in regulation, political priorities, training and awareness-raising initiatives.
▪ An EU directive on the public procurement of AI-based services and solutions in the public sector
A new directive or amendments to existing EU regulations should legally require member states to consider data ethics when procuring trustworthy AI for public sector purposes. The requirement could be integrated into all phases of the procurement process. Such integration should be guided by a due diligence approach, as illustrated by this white paper.
▪ Guidelines on the public procurement of AI-based services and solutions in the public sector
As an alternative or supplement to an EU directive, official EU guidelines should be adopted. The main purpose should be to guide the establishment of procurement and due diligence processes in public sector. It should illustrate how the inclusion of data ethics in the aforementioned processes may lead to procurement of trustworthy AI, including data ethics.
Also, these guidelines should inform private sector suppliers about requirements and expectations regarding their governance and management schemes in relation to design, development, deployment and maintenance of AI-based services and solutions.
▪ Inclusion of data ethics as a strategic policy priority in the Public Procurement Strategy of the EU
The EU Procurement strategy reflects the need to consider the environmental and social impact of new products and services as an integrated part of the procurement process. In light of the rapid development of AI and the expected influence on the decision-making processes of public administrations, it seems both timely and necessary to include data ethics as a new strategic priority when it comes to policy.
▪ A training toolkit on data ethics in the procurement of trustworthy AI-based services and solutions
To promote the principles of data ethics and improve knowledge and awareness of data ethics and due diligence in public procurement processes, a training toolkit should be developed.
It should address the public sector as the contracting party in the procurement process and the private sector in its role as the bidder on tenders issued by public entities.
▪ A data ethics handbook for the procurement of AI-based services and solutions
Complementary to the training toolkit, a handbook should be distributed. It should explain data ethics principles and how to integrate them when designing, developing, deploying and maintaining trustworthy AI-based services and solutions.
▪ An online help desk
If implemented as suggested, contracting public authorities and private sector tenderers
may need help understanding procedures, compliance requests, technical specifications
and organizational measures such as processes and the demonstration of the fulfilment of requirements. An online help desk would meet that need and be instrumental for greater, more active engagement in data ethics and trustworthy AI in the EU.