Parliaments are already exploring the integration of artificial intelligence (AI) technology for specific tasks. Reflecting on possible tools, application areas, usage scenarios, and requirements, it is reasonable to anticipate that AI-driven changes will manifest in parliamentary operations. Though Canada has been championing AI, additional research is necessary for its seamless integration and use in the parliamentary workspace. This research paper contributes to the bridging of this gap by presenting empirical evidence for the future use of AI-based tools and services, along with addressing open questions for their implementation within the Canadian Parliament. The data were collected during a brainstorming exercise in July 2020 and a virtual workshop in September 2023. An examination was conducted to investigate the relevance and priority of 210 applications and topics related to parliamentary AI.
involves interactive workshops, such as the ones organized in the Hellenic [
To address these research questions, an in-depth data analysis is initially conducted. The results of this analysis are then thoroughly evaluated by the researchers to draw meaningful insights and conclusions for the introduction and use of such technologies in the Canadian Parliament to prepare for the subsequent discussion with the participants of the Canadian House of Commons. This study, with data captured in September 2023, sheds light on the relevance and prioritization of AI-based technologies in the parliamentary workspace.
This paper represents a work in progress and serves as an intermediate research step. It is intended to lay the groundwork for subsequent discussions with participants from the Canadian Parliament, which will take place during a follow-up workshop. This intermediate step is considered important for refining the analytic framework before discussing and validating the results with parliamentary professionals. 2
Canada, a federation comprising ten provinces and three territories in North America, is a parliamentary democracy and operates as a constitutional monarchy in the Westminster tradition. The Parliament of Canada is the federal legislature parliament. It is seated at Parliament Hill in Ottawa and composed by the King, the Senate, and the House of Commons. The House of Commons has 338 elected Members of Parliament (MPs). The Senate as Upper House has 105 nominated senators. Legislation can be initiated by any MP or senator, and while the lower house is the most active in that regard, both must approve all bills. The governor general, on behalf of the monarch, provides royal assent to make bills into law. The official languages of the bilingual Parliament are English and French.
The Canadian Parliament uses information and communication technologies (ICT) that
facilitate and support its parliamentary functions. Within the administration of the House of
Commons, the Digital Services and Real Property Department is in charge for information
technology and information management. The IT team is mostly self-sufficient choosing not to
depend on outsourcing, but to develop own information systems, web apps and cloud
environment. Effectively, the IT team serves both chambers in their day-to-day IT operations. It
allocates the majority of its funds for operational solutions and innovative projects. The
operational focus of IT personnel makes the introduction of AI-based solutions even more
challenging. Moreover, the trust of citizens in Canada has been deeply shaken by several
dramatic cases of mismanagement in the government’s digital transformation procurement
processes [
Introducing innovative approaches, such as AI-driven predictive analytics for decision-making, AI-enabled constituent engagement, and AI-powered tools for legislative drafting, can significantly enhance parliamentary efficiency and responsiveness. These new perspectives are essential to address complex societal challenges, promote inclusive governance, and ensure that parliaments remain future-ready. By demonstrating these practical benefits for the particular case study of the Canadian Parliament, this ongoing research project can offer valuable insights and initiate meaningful change in parliamentary processes and societal dynamics.
Yet, positive response to workshop invitations and continued participation in this strategic
assessment demonstrates that the Parliament of Canada is committed to the increasing
utilization of AI which is growing in importance and cannot be overlooked. Lessons learned from
preliminary studies [
In late 2022, the introduction of ChatGPT [
The Parliament of Canada was also among these early adopters, contributing with various
applications, including those for closed captioning, automatic translation, physical and
cybersecurity, minute generation, summarization, and authentication [
The study's design primarily focused on identifying a suitable research approach that could
facilitate the evaluation of a wide, detailed, practical, and diverse range of potential application
areas for AI in parliaments, avoiding a one-sided perspective. It should be noted beforehand
that AI tools can be technically divided into several branches of operation such as
summarization, classification, sentiment analysis, semantic analysis, and recommendation.
Specific technologies and algorithms, for example Natural Language Processing (NLP),
Bidirectional Encoder Representations from Transformers (BERT), and Generative Pretrained
Transformers (GPT), can be applied differently depending on the case. Nevertheless, due to the
rapid evolution of technologies and algorithms, a technology-agnostic approach was
recommended. Additionally, the study should encompass more than just compiling existing
solutions. It should also involve capturing ideas for the future of parliament, even if they
currently seem technically unfeasible. A lot of these ideas are suitable as guiding pictures that
can be further developed into long-term visions and design-oriented approaches, while laying
the foundation for impact assessments [
The brainstorming method was chosen for the open collection of ideas [
The preliminary results of the brainstorming can be presented to the parliamentary
community for a more in-depth assessment. Rather than having a mixed participation from
different parliaments (see [
More than 3 years after the brainstorming exercise, a virtual workshop was held on September 11, 2023 in the Canadian Parliament. The aim was to evaluate the 210 proposals from the Canadian point of view. The 20 participants ensured a gender-balanced approached, that is nine men, ten women and one person that “preferred not to say”. The workshop included representatives from various sectors, including ten from IT, six managers from the Procedural Services (PS), the Office of the Law Clerk and Parliamentary Counsel (OLCPC), the Office of the Clerk and Secretariat (OCS), and the Digital Services and Real Property (DSRP), two from the legislative drafts, and each one from the library and the committee support. MPs and their aides were also invited to represent the demand side of parliament, but none of them choose to take part.
The preparation phase lasted two weeks during which several discussions were held to clarify various aspects of the study and specific organizational questions. To save workshop time, the participants were sent in advance the English and French translations of the original list of 210 proposals. Subsequently, the participants individually and anonymously assessed all 210 AIrelated proposals, each divided into blocks covering the nine thematic areas. The relevance and priority values for each of the proposals were captured and documented in the setup explained in the aforementioned section.
Concerning the outcomes derived from the Canadian Parliament workshop, the relevance
scores for all proposals ranged from 3.67 to 9.06, on a scale ranging from 0 to 10. The five highest
rated proposals (top 5) received a score of 8.57 or better. Just one out of 210 proposals (0.5%)
had a score of 9.0 or better, while 18 proposals (8.6%) were rated 8.0 points or better. The
cutoff-point of 7.5 and better, which is crucial for the relevance scale, contained 35 of 210 proposals
(16.6 %). 177 out of 210 proposals (84.2%) scored above 5.0 (relevant) and 210 (100%) scored
above 2.5. These results and the evaluation of the participants’ opinions underline a high
interest in AI for the future work operations of the Canadian Parliament. Overall, there was
valuable feedback for the proposals and, by extension, for the research and development
agenda of the research team. The recommendations for the implementation of these proposals
suggest an implementation interval between April 2024 and June 2029 [
To begin our analysis, we conducted two assessments on the comprehensive dataset. The initial compilation includes the top 10 proposals identified as the most relevant (Chapter 5.1, Table 1). This entails a focus on identifying projects what should be implemented. Secondly, the top 10 proposals rated with the highest priority (Chapter 5.2, Table 2) are put together to see which ones should be implemented first. The results of the evaluation of the participants' choices from the Canadian Parliament underline a high interest in AI-based technologies. The top 10 of all 210 proposals have received a relevance score of 8.25 or better on a scale from 0 to 10. The standard deviations (SD) range from 0.05 to 0.34. Only 18 of 210 values (8.6%) lies at 0.30 or above. Such large standard deviations indicate a rather divergent assessment within the group. 5,01 [5.-] AI-based translation services (texts) 4,01 [27.-] Intelligent, AI-based search functions in the front end
of the parliament's website 5,02 [6.-] AI-based translation services in real time for TV appearances/video appearances/video conference/ diplomacy/webinars/seminars/conferences (Babelfish) 1,01 [52.-] AI-based real-time subtitling of speeches by MPs in
parliament 3,01 [55.-] AI-based translation of parliamentarians', political 8.57 groups' and parliament's texts for TV/ radio/web/social media channels in the context of parliamentary diplomacy (GR/DE/EN/FR/RU/
AR/TR) and vice versa 5,03 [38.-] AI-based cybersecurity software 1,02 [53.-] AI-based real-time translation of speeches made by 8.48
minority members of parliament in their mother tongue 5,04 [71.-] AI-based automatic text and speech capture 8,01 [154.-] Clear accountability structures 5,05 [226.-] Virtual AI assistants for the disabled (e.g., reading and navigation aids) on the websites of the Parliament
There is only one requirement (8,01) in this list. The rest is filled by nine proposals (1,01; 1,02; 3,01; 4,01; 5,01-5,05). No open question is part of the top 10. An AI-based translation service for texts has received the highest value (9.60) and the lowest standard deviation (0.05). As Canada is a bilingual country, this brings in the steady need for fast subtitles and perfect translations from English to French and vice versa. AI captioning and AI translation services for texts, in real time even for subtitling speeches, for all audio and video formats, are in high demand. They facilitate multilingual interaction and increase mutual understanding. The ranking of this proposal is not surprising as an automatic translation service that uses the parliamentary knowledge base and a closed captioning solution already exist. Chapter 13 of the House of Commons Procedure and Practice allows debates to be conducted in languages other than English and French. Therefore, the need for AI-based real-time translation of speeches made by minority members of parliament in their mother tongue or by foreign guests in their mother language makes sense, too.
An intelligent, AI-based search function on the front end of the parliament's website is also highly rated, reflecting the need for helpful responses to user demands. The inclusion of virtual AI assistants in the parliamentary website designed to assist individuals with disabilities, offering functionalities such as reading and navigation aids, has received high approval. Considering the many increasingly sophisticated hacker attacks, the high demand for AI-based cybersecurity software that promises better protection is not surprising. A cybersecurity feature based on behavioral analysis is already in place in the Parliament of Canada, thus explaining the ranking for this proposal. The only framework requirement in the top 10 calls for a transparent accountability framework to ensure clarity, full responsibility and openness in the execution and oversight of AI activities. 5.2
Top 10 Priority of all Proposals With regard to priority, also to be related with implementation expectations, it can be observed that the participants in September 2023 set target dates in the years from April 2024 (value: 3.25) to June 2029 (value: 8.45). This corresponds to a period of five years and lies within a manageable planning horizon. The top 10 priority proposals should have been implemented by May 26, 2025 (value: 4.4) or earlier. The date of the next parliamentary elections in October 2025 might have played a role in making these choices. All standard deviations range from 0.13 to 0.29. None lies above 0.30, thus indicating a nondivergent assessment.
Among the top 10 proposals for the Canadian Parliament to be implemented as a matter of priority (until May 2025), there are seven concrete proposals (2,04; 3,01; 4,01; 5,01; 5,02; 5,04; 6,01), one requirement (8,02) and two open questions (9,04; 9,05). According to the evaluation, AI-supported translation services for texts should be implemented first. A similar solution is already in place, thus explaining the top priority position. An AI-based search function for the parliamentary website is also demanded to relieve burden if it provides a high-quality and rapid responses to queries. The question of whether MPs should generally have the option for remote access, enabled by technologies like 5G networks, is subject to various considerations. Factors such as virtual private network software, technological feasibility, security measures, representation concerns, and the nature of parliamentary proceedings, all contribute to the 5,01 [5.-] AI-based translation services (texts) 5,02 [6.-] AI-based translation services in real time for TV 8.65 0.20 19.07.2024 0.20 appearances/video appearances/video conference/diplomacy/webinars/seminars/conferences (Babelfish) Relevance 0..10 ↓Ø
SD 6,01 [223.-] Should MPs generally be allowed remote access (from a distance) to parliament and votes? (possible through 5G networks) 3,01 [55.-] AI-based translation of parliamentarians', political 8.57 0.14 05.02.2025 0.13 groups' and parliament's texts for TV/radio/web/social media channels in the context of parliamentary diplomacy (GR/DE/EN/FR/RU/AR/TR) and vice versa 2,04 [16.-] AI-based creation of texts and drafts based on 7.38 0.28 10.03.2025 0.23 parameters 9,04 [136.-] Ethical aspects of the operation of AI-based systems 7.45 0.25 19.04.2025 0.20 9,05 [97.-] Reflection on the limits of the use of AI in parliament 7.40 0.27 26.05.2025 0.20 8,02 [86.-] Expert opinion on the use and risks of AI in 8.05 0.17 26.05.2025 0.21 parliamentary work 5,04 [71.-] AI-based automatic text and speech capture 8.40 0.18 26.05.2025 0.24 complexity of this issue, where the group would like to have a solution soon. Generative AI features to automatically create text drafts on the foundation of drafts, specifications and other parameters can be offered by large language models. For reasons of national security, parliaments might opt here for own protected solutions to prevent external influence on politicians, parliamentary groups, and processes. Regarding the use and risks of AI in parliamentary work, the participants would like to rely on the opinions of recognized experts to reflect on these and to form their own opinion on the potential and dangers. In this context, two open questions need to be clarified soon, first the identifying and addressing of ethical aspects of the operation of AI-based systems in a parliament and second the limitations of AI use in the parliamentary proceedings.
This research promotes our understanding of the effects on AI-systems and AI-services in
parliaments. As such, it can contribute together with previous findings in the Hellenic and
Argentinian Parliament to the development of robust and comprehensive guardrails and
guidelines [
The novelty of this research is apparent. This is the first publication to strategically prioritize AI technology in the context of the Canadian Parliament. By focusing on this specific legislative body, the study offers insights and recommendations that have not been previously explored. Moreover, in a broader parliamentary context, these findings provide a valuable foundation for comparative analysis and broader discussions on AI utilization in parliamentary processes. The researchers have already initiated parallel workstreams to explore parliamentary AI integration in other parliaments, allowing for a comprehensive understanding of how such technologies can enhance parliamentary institutions functions.
By presenting a detailed case study of the Canadian Parliament, this research contributes to the existing body of knowledge by offering a particular parliament’s (and as such unique) perspective on AI implementation. It can thus serving as a critical reference point for future studies and policy development in the realm of AI and governance. This strategic prioritization not only underscores the potential of AI to revolutionize parliamentary operations but also sets the stage for ongoing and future comparative studies, thereby enriching the academic discourse on AI in the parliamentary context. 7
The Parliament of Canada is already utilizing AI technology within the parliamentary workspace. A reflection group investigated a number of constructive proposals for the use of AI in parliaments. The exchange and evaluation of the 210 proposals in terms of their relevance and priority has helped the IT team to form its own opinion on further fields of application and projects within the own parliamentary environment. The interactive workshop in September 2023 also constituted an important milestone in the internal discussion of AI, which will be followed by defining goals, acquiring budget and a step-by-step implementation of selected AI approaches. From the perspective of the workshop organizers, this goal-oriented approach with a clear interest should be emphasized positively. In contrast to earlier parliamentary workshops in 2021 and 2022, carried out before the rise of generative AI, the overall assessment can be characterized as more critical, more realistic, and not quite as overrated. This pragmatic approach could be contributed to the previous experience of the Parliament of Canada with AI. A comparative analysis is planned to be carried out in subsequent publications.
The chosen approach certainly has its limitations. Instead of starting a joint brainstorming session together with the participants on possible fields of application, a list of 210 proposals was used, which had already been developed in a previous workshop. In September 2023, this was particularly attractive for the Canadian team. Subsequently, they wanted to brainstorm their own ideas for possible concrete AI projects in an in-house brainstorming workshop. The aim of analyzing the extensive list was to help them to gain a comprehensive overview in advance so that they would not ignore any of the key opportunities. At this stage, however, no statements can be made on why exactly these approaches were selected and what experiences are already behind the motivations.
In a next step, to transfer these research findings into a Canadian case, the top three options for all eight topic clusters will be identified according to their relevance and then discussed with parliamentary staff. This will provide an overview across all thematic areas outlining the priorities set within each cluster. Finally, the top 3 open questions (sorted by relevance) should also be analyzed to determine which topics require attention and which institutional discussions need to be initiated as a matter of priority. In a further workshop, the results will then be reflected together with the Canadian team to understand and explain the background for the evaluation. A comparative analysis across the three parliaments is planned for the long term.
The evaluation results that also include a roadmap lie on the table. They provide valuable suggestions on where early commitment could make sense. However, it is up to those responsible in the IT team of the Parliament of Canada to decide which AI solutions to tackle next, which open questions to answer, which funds should be made available and whether implementation efforts should be made internally or with the help of external developers. These questions are not just about projects, but also about the right framework and the windows of opportunity in Canadian politics. The Parliament of Canada is preparing a major revision of privacy laws and, concurrently, is working on legislation specifically tailored to regulate AI effectively. This may lead to new tensions. Some legal issues, such as intellectual property rights and copyrights for training data, have not yet been conclusively clarified.
Waiting, however, is not always advisable. New projects with available AI services that are technologically feasible could be initiated for testing purposes. Some parliamentary end user groups (such as MPs, legal experts, or the parliamentary committees) might demand innovative AI solutions. In a very regulatory compliance focused environment, any algorithmic-based system would be under intense scrutiny from the very beginning. This raises the question of who will push and implement these changes in parliament. The push for innovation depends on the level of innovativeness of the leading group within any professional body. A team of leaders might be necessary for helping a community adopting a more aggressive AI approach. This is more difficult to organize in a parliament where people do not have the same level of digital competence and agility.
Canada grapples with an IT budget scandal due to procurement mismanagement around the
development of the ArriveCan app [
Reality will show which of the 210 proposals will gain real relevance and how quickly the Canadian Parliament will deal with them. Solutions already exist for some of the proposals, and it is realistic to expect that most of the top 10 relevant and top 10 priority proposals are currently assessed for potential implementation.
In the next years, it is essential for legislative bodies and external stakeholders, with academia taking a leading role, to engage in discussions regarding the utilization of AI within the parliamentary setting. These discussions should involve, among others, contentious debates about the boundaries of AI, the implementation of necessary regulations, and the collaborative design of solutions by the stakeholders themselves. With the presented results from the Canadian case, everyone involved can expect a considerable amount of work to advance their realization.