1 of 18

Artificial Intelligence in Public Procurement for Integrity and Efficiency

Paulo Magina

Head of Division,

Infrastructure and Public Procurement Division, OECD

Restricted Use - À usage restreint

2 of 18

2

Table of contents

1

2

Understanding AI in Public Procurement

The work of the OECD on AI and Public Procurement

3

Case studies

4

Future Trends

Restricted Use - À usage restreint

3 of 18

Understanding AI in Public Procurement

3

1

Restricted Use - À usage restreint

4 of 18

The digital transformation is an ongoing phenomenon since the 1970s

Digitisation: The converstion of analogue data and processes into digital formats.

Digitalisation: The use of digital technologies and data that result in new activities or changes to existing ones.

Digital Transformation: Encompasses the whole organisation, not just the processes, and typically involves cruss-cutting changes to strategy and behaviours. Use of AI is also changing the procurement landscape.

Restricted Use - À usage restreint

5 of 18

The type of use of innovative technologies in public procurement to strive to optimise performance and achieve better outcomes

Note: The figure presents data for respondent countries. Data for Denmark, Japan, and Switzerland is not available.

Source: OECD (2024), Survey on the OECD Recommendation on Public Procurement 2024

Measuring and enhancing efficiency are common goals for many countries, as they strive to optimise performance and achieve better outcomes.

Impact Assessments:

19 OECD countries measure administrative efficiency or have such measures under development.

Expanding the use of innovative technologies – AI, Robotic process Automation, cloud storage, etc. – can also drive greater efficiency.

Restricted Use - À usage restreint

6 of 18

Emerging technologies are used in all phases of the procurement lifecycle

AI has become an integral part of the public procurement lifecycle, basically impacting each step with its transformative capabilities.

Restricted Use - À usage restreint

7 of 18

Steps in the procurement cycle	Budget/spent analysis	Market study (with identification of potential suppliers)	Supplier management	Tendering phase	Contract management	Payments
AI application	Machine learning	AI-driven data analysis	AI-driven data analysis and data management	Bots	Natural language processing	Machine learning
Tasks to address	Classification of spend data into standard taxonomies.	Benchmarking suppliers and recommending new suppliers to be involved in market engagement activities.	Supplier risk monitoring based on real-time public data. Management of supplier register (creating and supplementing vendor data).	Virtual purchasing assistants for handling routine inquiries from bidders, e.g. to provide real-time updates to suppliers about the status of their bids.	Contract lifecycle management with reduced manual oversight and automated monitoring (e.g. compliance checking).	Payment tracking and verification (with identification of errors and fraud).

There is potential for AI applications throughout the public procurement cycle

Restricted Use - À usage restreint

8 of 18

Pros and Cons for AI in public procurement

Improved efficiency
Enhanced decision-making
Cost saving
Supporting the oversight of public procurement operations
Citizen engagement
Accessibility and inclusivity
Improving demand forecasting and strategic procurement planning, gaining insights into market trends

Ethical and bias concerns
Transparency and explainability
Data privacy and security
Lack of standardisation
Regulatory and legal frameworks
Safety and robustness
Public trust and perception
Resource intensiveness – especially data collection and management

CHALLENGES

OPPORTUNITIES

AI procurement is much more than a simple IT acquisition as it impacts data quality and continuity, challenges both dynamic and static systems, has unforeseen costs, and impact the whole organisational decision-making.

Restricted Use - À usage restreint

AI is poised to have a profound influence on public policies as it has the power to enhance government operations and the efficiency of decision-making, elevate communication and engagement with citizens, optimize speed and caliber, and lower the costs of service delivery.

AI technologies also pose challenges and risks, reflecting both the complexities of developing advanced AI systems and the ethical considerations associated with their deployment, ranging from ethical and bias concerns through regulatory gaps and lack of standardisation to resource intensiveness and risk-aversive behaviour of public entities.

The integration of AI into public procurement processes has become a focal point in the contemporary landscape of governance, and the more this technology infiltrates everyday lives, the greater the expectation is for AI technology adoption in the public sector and public services. However, governments are still relatively inexperienced in acquiring AI solutions and often grapple with issues related to the understanding of AI, legal and ethical considerations, and the need for skilled personnel to navigate the intricacies of AI procurement.

Opportunities for AI in public procurement:

Making data-driven decisions and increasing operational efficiency

New and more sophisticated digital solutions are being implemented to improve daily operations and decision-making. For example, the Australian state of New South Wales developed an AI tool that automatically categorises procurement transactions, replacing the manual categorisation of 2 million transactions per quarter.

Improving demand forecasting and strategic procurement planning, ensuring that procurement activities align with actual needs and budgets, and to gain insights into market trends.

Automated algorithms can compare bids, detect anomalies, and flag potential bid collusion or fraud.

As I mentioned, we are working with different oversight bodies such as the Portuguese SAI to apply AI and machine learning models to the identification of “red flags” associated with irregularities in public procurement processes. The project aims at increasing the reliability of the detection of irregularities and improving transparency and accountability.

Citizen engagement:

Chatbots and Virtual Assistants: AI-driven chatbots can provide citizens with instant, 24/7 assistance, answering queries and providing information.
Personalized Services: AI can enable the customization of public services, tailoring information and interactions to individual citizen needs.

Challenges for AI in public procurement:

Digitising procurement systems and processes is constantly evolving area, one of the challenges is to ensure that both the public and private sectors have the capacity to use these (ever changing) systems. Digital transformation cannot be successful without proper “change management” strategies.

AI relies on data to provide meaningful insights. This data can be challenging for public procurement agencies to collect and manage.

A robust investment in data collection, storage and data management systems should be integral part of digital transformation of public procurement.

With data collection and sharing increasing, it is also important to ensure data privacy and security. Procurement agencies need to adhere to strict data protection regulations and implement robust cybersecurity measures to safeguard sensitive information.

3. Regulatory and Legal Frameworks

Regulatory Gaps: Rapid advancements in AI have outpaced the development of comprehensive regulatory frameworks, leading to challenges in ensuring responsible and ethical AI use. In addition, the use of AI systems by the public sector across different administrative levels might lack the involvement of examination, oversight, or clear transparency.
Capacity issues: Institutional and government capacity limitations, both in the procurement and implementation phases, may result in discriminatory or faulty systems embedded in core public functions.
Regulatory Divide: Different national frameworks of procurement rules and regulations result in different approaches to the uptake of AI technologies in the public sector. Furthermore, coordinating AI regulations on a global scale presents difficulties due to varying cultural, legal, and ethical norms.

9 of 18

The work of the OECD on AI and Public Procurement

9

2

Restricted Use - À usage restreint

10 of 18

The OECD’s work on digital transformation

The OECD’s Recommendation on Public Procurement provides a platform for digital transformation that is integrated with broader public procurement goals

Direct support to countries to enhance the digitalisation of their public procurement systems
Thematic reports and policy papers to identify trends and promote good practices

The OECD Recommendation on Public Procurement

Restricted Use - À usage restreint

The OECD’s work is based on the 2015 OECD Recommendation on Public Procurement, comprised of 12 integrated principles. “Integrated” because the principles are interlinked and support each other’s implementation.

One principle of this Recommendation, e-procurement, is dedicated to the “The digitalisation of procurement processes”. However, there are close links with many other principles. To cite some of them:

Transparency and Integrity: E-procurement tools facilitate the access to public tenders and improve transparency of public procurement processes as well as accountability of procurement officials, which contributes to mitigate risks of corruption inherent to procurement processes.

Evaluation: Digitalisation enables the collection and use data which is necessary for the assessment of the performance of procurement processes and outcomes. Governments can better measure and understand the impacts of their purchasing decisions, as well as better understand and shape the markets from which they are buying.

Participation: Digitalisation enables for more opportunities for stakeholders participation in the public procurement system (with suppliers, business associations, citizens, NGOs etc.).

Interoperability:

We work closely with OECD colleagues who specialise in digital government and AI to bring multiple perspectives and areas of expertise (OECD AI Principles, Recommendation on Digital Government Strategies)

We provide direct support to several countries to enhance the digitalisation of their public procurement systems (e.g., Ireland, Mexico, Korea, Portugal, Tunisia), working with diverse stakeholders, such as procurement policy makers, CPBs and oversight bodies. We also look at digitisation as one area of comprehensive public procurement reviews (e.g., MAPS, Germany).

Under our current Programme of Work and Budget, we are developing good practice reports that will serve as useful resources for countries:

One report will provide evidence on the use of digital tools and data to support public procurement through a multi-dimensional approach. This will include the way public procurement systems are collecting and using data to measure and manage performance and impact. It will also provide information about the integration between digital procurement and other systems.

Another planned good practice report will focus on public procurement mechanisms enabling the implementation of digital government strategies. This will look at good practices on how public procurement can support a successful digital transformation of governments and public institutions under the principles of agility, horizontal collaboration and integration responding to user needs. The aim will be to identify specific areas to further align public procurement and digital government policies to enable an effective and sustainable digital transformation of the public sector.

11 of 18

OECD definition of an AI system – The OECD AI principles (2019/2024)

An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.

Source: OECD AI Principles

Restricted Use - À usage restreint

12 of 18

Expert Group on AI and Public Procurement

At the 2024 meeting of the Working Party of Leading Practitioners on Public Procurement (LPP), delegates highlighted diverse approaches to integrating AI into public procurement systems globally and opted for leveraging the use cases and good practices on AI and procurement in the form of an expert group to solicit expert feedback to set the path for future work.

The Expert Group has over 30 participants from 12 OECD member and partner countries and 4 intergovernmental organizations.

Bimonthly meetings. Themes include:

Integrating AI into the procurement cycle and using AI to inform public procurement strategies
Training and capacity building for AI
Guidance and frameworks for the procurement of AI

The overall objective is to strengthen the collaborative efforts to support governments in establishing a nourishing and safe policy environment for the implementation of AI in public services.

A paper on the procurement of AI is planned to be published in the third quarter of 2025. The outputs of this Expert Group are also expected to feed into the discussion on emerging technologies in public procurement at the Global Public Procurement Forum (1-2 July 2025, Paris, OECD Headquarters).

Restricted Use - À usage restreint

13 of 18

Case Studies

13

3

Restricted Use - À usage restreint

14 of 18

The case of Portugal: Enhancing Control Oversight at Tribunal de Contas

Enhancing efficiency and transparency of public procurement by strengthening the overall control framework through developing data governance strategies and a machine learning tool.

The project's key focus was to leverage artificial intelligence (AI) and machine learning to enhance the TdC's capacity to identify and assess risks and irregularities in public procurement.

Long-term

Short/Medium term

Increased reliability in detection of irregularities/risks related to public procurement
Improved public procurement transparency, accountability and efficiency of the use of public funds
Accelerated digital transformation of the TdC.

Better data governance and improved use of available PP information
Stronger control capabilities and more efficient allocation of resources by the TdC

Main outputs :

OECD report on “the efficient financial compliance and control of public procurement”
Development of a data-driven risk-based model (with NOVA IMS)
Guidelines and training workshops for auditors
Opening Conference (Jan.2023) and Final Conference (Feb.2025) in Lisbon

Restricted Use - À usage restreint

The project – funded by the EU TSI – aims to enhance the efficiency and transparency of public procurement by strengthening the overall control framework. The TdC, as part of its mandate, conducts thousands of audits annually, including a priori, real-time, and posterior audits of public procurement activities. These audits are essential but resource-intensive, requiring significant human and financial investments. The OECD's support has been instrumental in developing strategies to streamline these processes and improve their effectiveness.

The project's key focus has been leveraging artificial intelligence (AI) and machine learning to enhance the TdC's capacity to identify and assess risks and irregularities in public procurement. By integrating these advanced technologies, the TdC aims to automate and optimise the audit process, making it more efficient and accurate. The use of AI and machine learning allows for the analysis of vast amounts of procurement data, identifying patterns and anomalies that may indicate potential issues. This proactive approach not only improves the detection of irregularities but also helps in preventing them, thereby strengthening the overall control framework.

15 of 18

The case of Finland: AI advises on public procurement

The Ministry of Finance has introduced an AI chat bot in the Hilma public procurement service. The bot is intended to assist experts in public procurement with related questions.

The bot can practically answer any question about procurement. The tool also informs users if it does not know the answer to a question.

The benefit of the AI chat bot lies in the complexity of public procurement and the fragmented nature of related information. Public procurement can be a complex and confusing subject for people who do not yet know much about it.

The Ministry of Finance has also identified some problems with the bot's implementation as it can also hallucinate and give incorrect answers. Therefore, the chot bot does not yet provide official advice.

Restricted Use - À usage restreint

16 of 18

Future trends

16

4

Restricted Use - À usage restreint

17 of 18

AI is changing the procurement landscape

Public Procurement as Facilitator and Gatekeeper

Facilitator: Public procurement can drive innovation by adopting AI technologies, enabling efficient and effective service delivery.

Gatekeeper: Ensures responsible AI adoption by setting standards for transparency, trustworthiness, and ethical use.

Impact on Professionalisation

Skill Enhancement: AI tools require procurement officers to develop new skills in data analysis, AI management, and strategic decision-making

Role Evolution: Shift from manual tasks to strategic roles, focusing on AI oversight and implementation

Efficiency and Transparency

Operational Efficiency: AI automates routine tasks like supplier selection, invoice processing, and contract management, reducing time and costs

Enhanced Transparency: AI-driven analytics provide real-time insights, improving decision-making and accountability

Restricted Use - À usage restreint

18 of 18

DRAFT / UNDER DEVELOPMENT

18

THANK YOU!

Paulo Magina (paulo.magina@oecd.org)

Restricted Use - À usage restreint