International Collaboration and Examples

While legislatures around the world often deal with very different political dynamics and issues, the adoption of new technologies is one area in which these institutions can benefit from sharing information and learning from one another. The Inter-Parliamentary Union has been an important convener and information hub for the digital transformation of parliaments worldwide, especially through the work of its Centre for Innovation in Parliament and resources such as the recent guide to digital transformation:¹

In reflecting on her research into AI in public institutions: Deputy Chief Information Officer (CIO) of Brazil’s Chamber of Deputies Patricia Almeida described three stages of AI adoption in parliaments:

  1. The first group includes parliaments that have already implemented AI projects successfully and are now focusing on making these solutions sustainable. Their focus is on policy and strategy development and creating requirements for public procurement that include AI services.

  2. The second group includes parliaments that are experimenting with AI in a “lab mode” — testing solutions to demonstrate their benefits and gain support and funding from parliamentary leadership.

  3. The third group consists of parliaments that have not yet adopted AI, primarily due to the lack of basic IT infrastructure, legislative systems, and data governance.

“Each group has different needs and can benefit from guidance and support, especially from inter‑parliamentary networking.”²

A 2020 article on “Artificial Intelligence in the Parliamentary Context” found, at the time, that only 10% of parliaments had adopted any kind of AI technology, with 6% related to legislative drafting. However, 45% of those surveyed indicated that they anticipated some kind of AI adoption in the two years that followed.³

Source: “Artificial Intelligence in the Parliamentary Context,” Techno-Politics Series: 4 — Smart Parliaments Data-Driven Democracy (2022)

In July 2023, The Open Government Institute (TOGI) at Zeppelin University, in partnership with the Hellenic OCR team, orchestrated an international research workshop on "AI in Parliaments" in Friedrichshafen, Germany. The event fostered a platform for representatives from diverse parliaments to exchange insights on AI policies and implementations in their respective countries, laying a foundation for potential future collaborations.

During a June 2023 workshop titled “AI for Parliaments” at Zeppelin University, participants from ten legislative bodies shared and discussed their AI endeavors.

In September 2023, three hundred parliamentarians, experts, and practitioners representing seventy parliaments from around the world gathered in Montevideo, Uruguay for the Second World Summit of the Committees of the Future, hosted by Parliament of Uruguay and the Inter-Parliamentary Union (IPU). Attendees discussed rapid developments in AI and challenges for legislatures worldwide. The outcome document emphasizes foresight, ethics, human rights, and global cooperation as central to government’s role in steering AI's evolution in line with democratic values. Discussions encouraged further opportunities for parliamentary exchange and coordination on AI, such as an online map of AI-related actions. Annual Committees of the Future summits are proposed to continue collaboration between global parliaments on AI governance.⁴

In September 2023, the “Second World Summit of the Committees of the Future” convened in Montevideo, Uruguay to discuss artificial intelligence in future committee work.

The following examples are a sample of the experimentation and advances being made in legislative bodies around the world:

Argentina

The Argentinian Chamber of Deputies’ “DipLab'' began experimenting with AI in 2019 — identifying processes that could benefit from AI, conducting training sessions, developing AI diagnostic tools with international partners, and creating an AI guide for parliamentary work.⁵ Existing AI projects include digitizing transcripts of plenary sessions from 2005 to 2020 and creating an editable transcript database in collaboration with the National Council for Scientific and Technical Research (CONICET). The algorithm indexes legislative information, allowing for effective searching and analysis of public policy trends and legislative changes. Future projects include developing a virtual tool for bill drafting with AI integration, digitizing all parliamentary archives, and launching additional AI-driven projects to improve efficiency and transparency in parliamentary work.⁶

Brazil

In Brazil, the Chamber of Deputies uses its ML-powered “Ulysses” platform to analyze and tag large volumes of documents and data and match content to user interests. The platform allows the public to express support or opposition and comment on pending bills (sometimes receiving up to 30,000 comments on one bill) and uses ML and NLP to analyze the comments.⁷ In exploring further public participation, the legislature is piloting AI chatbots to provide a conversational interface for citizens to access information on legislators, bills, and budget data. The Chamber is also experimenting with computer vision techniques like facial recognition to identify legislators during video broadcasts and plenary sessions.⁸

Across the legislature, the Brazilian Senate is exploring AI-based speech recognition services for various parliamentary activities.⁹

Estonia

The Estonian Parliament, the Riigikogu, has incorporated AI into its operations, using a system named HANS, based on large language models, to transcribe all parliamentary meetings, including plenaries and committees, through automated speech recognition. For plenary sessions, the AI-generated verbatim text is reviewed and corrected by editors before being published online. In committees, the digitally generated transcript is used by staff to manually prepare minutes. HANS has an error rate of around 5%, which can vary depending on the speakers and the meeting context. This implementation of AI has led to a reduction in staff, particularly in replacing the Parliament’s team of stenographers, who were nearing retirement and difficult to replace due to the nature of the job.¹⁰

Looking ahead, the Riigikogu plans to explore more advanced AI applications, such as generating automated summaries of committee meetings. To address concerns about reliability and accuracy, the Parliament is adopting a dual approach: developing technology in “lab mode” to understand its capabilities and limitations while simultaneously engaging leaders and politicians in discussions about AI requirements. Estonia is prioritizing human oversight, ensuring that committee reporters will review, edit, and finalize the AI-generated text, mitigating fears about AI replacing human judgment.¹¹

Screen-capture of the Riigikogu HANS interface displaying AI-generated transcription text

European Union

At the 2023 "AI for Parliaments" conference in Friedrichafen, Germany, Charalampos Moschopoulos, the EU's Project Leader in Artificial Intelligence, discussed the European Parliament's (EP) integration of AI in its administrative and operational processes. The EP has adopted various AI-enabled tools to enhance efficiency and effectiveness including chatbots, which automate question-and-answer processes across multiple sectors, aiding in quick and efficient responses to queries.¹² Additionally, the EP uses an auto-summarizer and Microsoft Editor to provide concise text summaries, facilitating the understanding of complex documents.¹³

The EP has also implemented eTranslation, a tool capable of translating documents into multiple languages, an essential function for a multilingual institution like the EP.¹⁴ Another significant integration is the Speech to Text service, which offers real-time transcription, enabling accurate and immediate documentation of spoken words during meetings and discussions. Furthermore, the EP uses its EUROVOC system for automatic document indexing within its system, streamlining the categorization and retrieval of documents.

Finland

The Finnish Parliament's Committee for the Future, established in 1993, engaged in a unique experiment as part of its foresight activities. In 2021, the committee conducted a hearing using simulated personalities from "Project December," an AI project enabling conversations with chatbots modeled after real or fictional characters. This hearing, which was structured similarly to standard parliamentary hearings, involved a detailed preparation process. During the session, the committee members interacted with two AI personalities, Muskie and Saara, posing questions and receiving written responses.

This initiative's purpose was to explore the potential of AI in providing perspectives and information, and the technology’s ability to formulate arguments and counterarguments, akin to the contributions of human experts. After each interaction, the committee members discussed the AI responses and their implications with the facilitators. This innovative approach by the Finnish Parliament's Committee for the Future highlights their commitment to integrating advanced technology into legislative processes and exploring AI's role in enhancing parliamentary functions.¹⁵

India

In India, a newly inaugurated parliament building incorporates several AI-powered technologies to enhance operations, accessibility, and security. One innovation is the “Digital Sansad” application¹⁶ that allows Members of Parliament to listen to proceedings in their own languages in real time. This AI-enabled platform leverages automatic speech recognition to accurately transcribe and translate speeches word-for-word. It aims to facilitate participation and understanding across India's linguistic diversity.

The building’s security also utilizes AI, including biometric scanning and facial recognition to securely identify and grant access to officials. As MPs approach the entrance gates, scanning technology recognizes and verifies their identity, automatically opening the gates without need for badges or cards. Inside, access controls regulated by AI limit entry to sensitive areas like the Prime Minister's wing.¹⁷

Italy

The Italian Senate utilizes an AI system, “Gestore Emendamenti” (GEM), to manage amendments.¹⁸ The system employs text clustering algorithms to quickly identify groups of similarly worded amendments. “The aim is to assist the Senate staff in the detection of groups of amendments similar in their textual formulation in order to schedule their simultaneous voting.”¹⁹ While the system produces near-instant results, staff members review, approve, modify, and integrate the machine-generated analysis to ensure that no decisions are made without human oversight.

Singapore

Similar to the US Congress, the Ministry of Communications and Information (MCI) of Singapore issued guidelines for public officers and Members of Parliament on the use of AI tools like ChatGPT for research and writing in May 2023, allowing for responsible experimentation.²⁰ The guidelines explicitly instruct officers not to input sensitive information into these GenAI applications and require that all AI-generated work be reviewed to ensure accuracy and compliance with copyright laws. These guidelines target both general users of AI applications and developers creating apps for the government.

Singapore also released “Pair,” a suite of tools designed to amplify the productivity of public officers, including legislators. Pair offers a secure platform on government-issued laptops, allowing officers to harness LLMs' capabilities without compromising confidential data. Endorsed by the AI Policy Group within the Smart Nation and Digital Government Group, Pair operates on a question-answer mechanism, akin to a secure and swift version of ChatGPT tailored for public officers. It has received clearance from Singapore's Smart Nation Digital Government Office for use with documents up to the Restricted/Sensitive Normal classification, and its "bring your own data" feature allows officers to integrate personal datasets, enhancing its adaptability and relevance.²¹

Taiwan

Beginning in 2015, Taiwan's legislature integrated AI into its democratic processes through the vTaiwan (virtual Taiwan) and Pol.is platforms.²² vTaiwan is an online platform designed for public consultation and participatory legislation. Pol.is, an AI-driven tool, plays a crucial role in the vTaiwan process — participants can express their opinions on specific issues, and the AI system visualizes these opinions, highlighting areas of consensus and disagreement. This visualization helps legislators and policymakers to understand public sentiment on various issues in a nuanced and detailed manner. This innovative approach to lawmaking has been applied to various issues, including the regulation of online alcohol sales, telemedicine, and fintech regulations.

Despite the technical success of the initiative, the substantive success of deliberative fora for public input is still dependent on the acceptance of that input by policymakers. As World Bank governance expert, Tiago Peixoto, explained to The Daily Beast, “Technology made it extremely easy for any government to put something out there and ask citizens... but it didn’t make it easier for governments to be able to respond.”²³

Greater Collaboration and Knowledge Sharing

Innovation in AI adoption in parliaments worldwide presents a significant opportunity for legislative bodies to learn from each other as they adapt to and experiment with new technologies. International fora, such as the Second World Summit of the Committees of the Future and the “AI in Parliaments” workshop, have already begun to facilitate this exchange of knowledge and strategies. These events have proven valuable in showcasing the varied applications of AI in legislative processes, from Argentina's digitization of transcripts to Estonia's AI-based transcription systems, and from Finland's AI chatbot experiments in committee hearings to India's use of AI for real-time language translation in its new parliament building.

As legislative bodies continue to navigate the early stages of AI experimentation, the opportunity to learn from each other's successes and challenges becomes increasingly important. With the ongoing advancements in large language models and AI tools, and the growing interest in AI applications within legislative contexts, there is an expectation of more comprehensive AI integration in parliaments in 2024 and beyond. This trend underscores the potential for greater collaboration and shared learning among legislatures globally as they explore the capabilities and implications of AI in governance.


¹ “Guide to digital transformation in parliaments,” Inter-Parliamentary Union (2023)

² “Expert perspectives on AI in parliament,” Inter-Parliamentary Union Innovation Tracker (October 21, 2023)

³ Xenia Ziouvelou, George Giannakopoulos, and Vassilis Giannakopoulos, “Artificial Intelligence in the Parliamentary Context,” Techno-Politics Series: 4 — Smart Parliaments Data-Driven Democracy (2022)

⁴ “Second World Summit of the Committees of the FutureInter-Parliamentary Union (September 2023)

⁵ “Inteligencia Artificial, Proyectos del DipLab,” Diplab, Diputados Argentina

⁶ “Argentina: First steps towards an AI driven Chamber of Deputies,” IPU Innovation Tracker Issue 16 (October 21, 2023)

⁷ “Case Study: Brazil: A digitally mature parliament,” Inter-Parliamentary Union (January 6, 2022)

⁸ “Brazil: Ulysses machine learning for transparency and citizen participation,” Inter-Parliamentary Union (July 21, 2021)

⁹ F. Fitsilis, , & J. von Lucke, “Research and Development Agenda for the Use of AI in Parliaments,” The Journal of Legislative Studies (2022)

¹⁰ “A view from Estonia: AI for advanced beginners,” IPU Innovation Tracker Issue 16 (October 21, 2023)

¹¹ “Estonian parliament uses speech recognition technology to create verbatim records,” e-Estonia (September 2020)

¹² “Bringing the European Parliament closer to citizens through AI,” Microsoft (Sep 15, 2020)

¹³ F. Fitsilis, & J. von Lucke, “Research and Development Agenda for the Use of AI in Parliaments,” The Journal of Legislative Studies (2022)

¹⁴ Gregorio Sorgi, “EU gives more power to AI translation machines,” Politico (June 15, 2023)

¹⁵ Fotios Fitsilis, “Artificial Intelligence (AI) in parliaments – preliminary analysis of the Eduskunta experiment,” The Journal of Legislative Studies, 27:4, 621-633 (2021)

¹⁶ “Digital Sansad app: An AI-powered platform to transcribe house proceedings at the new Parliament building,” IndiaAI (June 1, 2023)

¹⁷ Milin Stanly, “The Indian government is leveraging the power of AI in Parliament building for safe and secure working,” IndiaAI (August 10, 2023)

¹⁸ “How AI helps the Italian Senate manage amendments,” Inter-Parliamentary Union Innovation Tracker Volume 12 (October 6, 2022)

¹⁹ Tommaso Agnoloni, Carlo Marchetti, Roberto Battistoni, Giuseppe Briotti, “Clustering Similar Amendments at the Italian Senate” (June 20, 2022)

²⁰ Osmond Chia, “Public officers can use ChatGPT and similar AI, but must take responsibility for their work: MCI,” The Strait Times (May 23, 2023)

²¹ “Pair - A Helpful Assistant Powered by Large Language Models (LLMs),” Singapore Government Developer Portal (September 29, 2023)

²² Carl Miller, “How Taiwan’s ‘civic hackers’ helped find a new way to run the country,” The Guardian (September 27, 2020)

²³ Nick Aspinwall, “Taiwan Tried to Digitize Democracy. It Was a Huge Flop,” The Daily Beast (November 1, 2023)

Previous
Previous

GenAI in the US Congress and State Legislatures

Next
Next

Strategic Considerations for AI Integration