Guidelines for AI procurement in government
Supporting the public sector in exploring artificial intelligence solutions
The World Economic Forum’s Center for the Fourth Industrial Revolution, in partnership with Deloitte, the UK government’s Office for Artificial Intelligence (AI), and other contributors, has developed guidelines for the public sector’s procurement of AI solutions, helping governments build leading procurement practices that can set new standards for efficient and ethical procurement and use of AI, and accelerate AI adoption in the public sector.
Leveraging procurement to drive public-sector AI adoption
Artificial intelligence holds great potential for public-sector institutions around the world to improve government operations as well as service to citizens. But governments don’t necessarily have experience in acquiring modern AI solutions and can tend to be cautious about harnessing new technology.
By helping to guide the process of procuring AI, we aim to address major AI adoption pain points early in the process and make it easier for governments to implement this advanced technology. Our goal is to help public-sector institutions to:
- Create clear specifications that address concerns of bias, privacy, accountability, transparency, and proportionality.
- Build dynamic, competitive AI marketplaces that allow governments to tap into the best talent.
- Build on established principles of good government technology procurement, including existing measures such as frameworks, model contracts, and diversity and inclusion practices.
- Establish responsible AI practices today so they’re prepared to assess the benefits and risks of ever-evolving future AI applications.
- Set a baseline for discussions surrounding AI standards in the market, which are often heavily influenced by a government’s moral authority and credibility, as well as its considerable purchasing power.
Overall, the guidelines aim to assist all parties involved in the procurement life cycle – policy officials, procurement officials and government commercial teams, data practitioners, and AI-solutions providers – in safeguarding public benefit and well-being.
Overview of the guidelines
The guidelines contain 10 high-level recommendations, which are discussed in detail in the full report . Briefly, we recommend that public sector procurement and commercial teams:
- Focus on outlining the problems and opportunities they’re looking to solve, and use procurement vehicles that promote iterative solution development.
- Define the public benefit of using AI, while assessing risks.
- Align with relevant existing government strategies and contribute to their improvement.
- Incorporate relevant legislation and codes of practice.
- Articulate the feasibility of accessing data that may be relevant for the AI solution.
- Highlight any limitations of potential data.
- Work with a diverse, multidisciplinary team.
- Focus on algorithmic accountability and transparency.
- Plan for the support that will be required to operate the tool throughout its life cycle.
- Create a fair, level playing field among AI-solution providers.
