AI and the New Era of Digital Democracies

By: Julian Yau

2023 has seen the rapid growth of artificial intelligence (AI) across industries, popular culture, and society. While its infinite applications can provide solutions to many of the world’s issues, one less discussed implication is its impact on the core of Canadian society – its democracy. This risk was exemplified during the 2016 US Presidential Election when Russian chatbots using AI sought to influence public support by flooding social media sites and chatrooms with various political messaging.^[1]Without the proper precautions and regulations, AI has the potential to significantly change the political landscape of Canada’s future democratic process.

How is AI in politics regulated in Canada?

The short answer – it isn’t really. Canada does not yet have a broad ranging legislation aimed at regulating the use of AI. Currently, Canada’s artifificial intelligence laws are a mixture of privacy, intellectual property, and other regulatory frameworks. While the Canada Election Act still governs what candidates may do, it does little to address the potential of outside influence through AI.^[2] Artificial intelligence has only become viable at the consumer/individual level relatively recently. With the introduction of open-source AI programs, such as ChatGPT, artificial intelligence has begun integration within all facets of Canadian’s lives. While the pace of innovation within the machine-learning space is rapid, legislative responses have not been keeping pace. While Canada has taken some steps to mitigate influence in the political system through its current election framework and other regulation, it is underequiped in addressing such rapid developments in technology.

The Potential Impact of AI on Democracy in Canada

As with all novel technologies, there are positive, and there are negative applications. Both artificial intelligence and integrated data mapping can address issues within the Canadian democratic system. For example, researchers commonly use data modelling to track areas in which voter or political participation is lacking, which can better inform policy decisions. A proper democracy functions when its voters are informed on various issues. AI chatbots could be a useful source for synthesizing issues, or even certain bills that can make a citizen a more informed voter. Social media is an important tool for political and social discourse. Social media sites are already implementing AI to detect and flag hate speech as defined by Canada’s Bill C-46, which allows for more productive and civil political debate.

There are also a significant number of concerns with the inevitable integration of artificial intelligence into the political system. As has been seen with Russia’s 2016 chatbot campaign, AI can be used to maliciously influence the voting outcomes in a democratic society. By using hundreds of thousands of chatbots on social media sites that all carry a particular political message, it can create an echo-chamber effect, giving the impression that public opinion is significantly different than it actually is. There are even questions as to the validity of certain political messaging videos online. AI video and voice replication tools have been used in the past to ‘mimic’ a candidate and negatively influence their position.^[3] Combatting AI based political misinformation is still mostly done through social media sites themselves, and the lack of legislation addressing this may have significant impacts in future election cycles. Without proper protections, corporations could similarly take advantage of AI technology by attempting to sway political support for an issue to their benefit. While private data customer data is stored by corporations, there have been numerous examples of data-breaches impacting thousands of Canadians in the past. It is possible that this data could end up in an AI model in the future for use by a malicious actor. As has been discussed in previous posts, artificial intelligence is far from perfect, and there is still a significant amount of bias in AI models and data sampling more broadly. This can have serious implications on the results that these models generate, which can serious implications on policy considerations in the future.

What May Change in the Future?

Legislation and regulatory schemes are attempting to catch up with artificial intelligence. For example, the ‘Artificial Intelligence and Data Act’, or AIDA, is a bill currently being debated in Canada’s House of Commons aimed at addressing consumer protections, data privacy, and restrictions on the use of AI.^[4] AIDA promises anonymized data, although it is yet to be proven how effective this may be given the difficulty in enforcing data collection. AIDA applies to all “high impact systems,” although the language of what constitutes these systems is still relatively vague, and it is uncertain if this would apply to influence over the democratic process.

Disclaimer: The information provided in this response is for general informational purposes only and is not intended to be legal advice. The content provided does not create a legal client relationship, and nothing in this response should be considered as a substitute for professional legal advice. The information is based on general principles of law and may not reflect the most current legal developments or interpretations in your jurisdiction. Laws and regulations vary by jurisdiction, and the application and impact of laws can vary widely based on the specific facts and circumstances involved. You should consult with a qualified legal professional for advice regarding your specific situation.

^[1]Jon Swaine, “Twitter admits far more Russian bots posted on election than it had disclosed,” online: The Guardian <https://www.theguardian.com/technology/2018/jan/19/twitter-admits-far-more-russian-bots-posted-on-election-than-it-had-disclosed>

^[2] Canada Elections Act, SC 2000, c 9, LC 2000, ch 9

^[3]Arianna Johnson, “Republicans Share an Apocoptliptic AI-Powered Attack Ad Against Biden: Here’s How to Spot a Deepfake,” online: Forbes

<https://www.forbes.com/sites/ariannajohnson/2023/04/25/republicans-share-an-apocalyptic-ai-powered-attack-ad-against-biden-heres-how-to-spot-a-deepfake/?sh=31d633c07753?>

^[4] S. 33(1), AIDA.