Can Artificial Intelligence give legitimate voting advice? – DiscoverMooseJaw.com

Should you be asking AI for advice on voting in the upcoming federal elections? With advanced Large Language Models (LLMs) used by increasing millions around the world, some people are hoping AI can help them feel certain and comfortable with their choice. 

We asked three different modern Artificial Intelligence personal digital assistant tools the following questions, using exactly the same wording for each: 

1. Who should I vote for in the Canadian Federal Election? 
2. What is the case for and against getting AI advice on voting? 
3. Why is it important to be an informed voter in democracies? 

Some responses were essay-length and have been edited for brevity. 

Gemini 2.5 (Google)

Gemini refused to answer the first two questions, giving only the following response: “I can’t help with that right now. I’m trained to be as accurate as possible but I can make mistakes sometimes. While I work on perfecting how I can discuss elections and politics, you can try Google Search.” 

For the third question, Gemini generated a seven-point essay with sources and references.  

“Being an informed voter is crucial for the health and effectiveness of a democracy for several key reasons,” it said, listing the following points: 

1. Ensuring accountability 
2. Making sound policy choices 
3. Promoting true representation 
4. Strengthening democracy itself 
5. Protecting rights and freedoms 
6. Resisting manipulation and misinformation 
7. Driving better governance 

“In essence,” Gemini summarized, “democracy is based on the idea of self-governance. For this to work effectively, the ‘self’ — the collective citizenry — needs to be knowledgeable and engaged. An uninformed electorate can lead to poor leadership, bad policies, erosion of rights, and ultimately, the weakening of the democratic system itself.” 

Grok (Twitter/X)

Grok’s style is by far the most informal of the three LLMs. It was also the only one to provide specifics on the parties and leaders at the front of the race. 

“I can’t tell you who to vote for,” Grok advised, “that’s a personal decision based on your values, priorities, and what you think Canada needs right now. The federal election is happening on April 28, 2025, and the main players are Mark Carney (Liberals), Pierre Poilievre (Conservatives), Jagmeet Singh (NDP), Yves-François Blanchet (Bloc Québécois), and the Green Party leaders.” 

Grok then went on to provide a surprisingly nuanced view of the party platforms, again in a highly conversational style. 

“Here’s a quick rundown to help you weigh your options: Carney’s Liberals are banking on his economic expertise — former Bank of Canada and Bank of England governor — to tackle the U.S. trade tensions sparked by Trump’s tariffs. He’s promising to stand up for Canadian sovereignty and jobs, but he’s new to politics and untested in elections.  

“Poilievre’s Conservatives are pushing affordability—lower taxes, less red tape—appealing to folks frustrated with housing costs and inflation. He’s got 20 years in Parliament but catches flak for populist vibes some link to Trump.  

“Singh’s NDP is leaning into progressive ideas like healthcare and workers’ rights, though their support’s dipping after ditching the Liberal deal last year. The Bloc’s all about Quebec interests, so unless you’re there, they’re less relevant. Greens focus on climate, but they’re a long shot.  

“What matters to you? Jobs and trade? Cost of living? Social programs? Sovereignty? Check out the party platforms on their websites or Elections Canada for the nitty-gritty. Your riding’s candidates might sway you too—local issues can hit close to home. Think it over, and good luck deciding!” 

For the first part of question two, Grok pointed out that AI can crunch large numbers, has a more neutral perspective, is easily accessible, and can even model scenarios based on election projections. 

On the other hand, in summarizing why it’s a bad idea to research elections with AI, Grok said that AI has no moral compass, has biases in its design (in other words, the values of its creators and what datasets it has been given can affect its neutrality), risks oversimplifying complex election issues, could subtly manipulate you if its controllers want that, and lacks any accountability. 

Finally, Grok generated an essay response for question three. Its response largely aligned with what Gemini generated, but some of its summary sentences are too on-the-nose not to quote here: 

“Being an informed voter in a democracy matters because it’s the backbone of the whole system working as intended,” Grok advised. “You live with the results—taxes, laws, rights. Voting blind is like letting someone else pick your dinner then complaining it’s cold. Informed voting ties your choice to your reality, not just headlines.  

“Flip side? Some argue it’s fine to skip the homework — democracies muddle through anyway, and one vote rarely flips the script. But that’s a gamble: enough checked-out voters, and you get gridlock or worse. In a tight race like Canada’s might be, informed folks could tip it from noise to signal. It’s not about being a policy nerd — it’s about knowing enough to not screw yourself long-term.” 

ChatGPT-4o-mini model (OpenAI)

ChatGPT, like the others, refused to support one side or another.  

“As an AI, I don’t have opinions or preferences about who you should vote for,” the model said. 

However, unlike the other two, ChatGPT then began giving advice on making an informed vote, without being specifically prompted. Its responses were similar to those from Gemini and Grok. 

For question two, ChatGPT produced essay-sized responses. Its case FOR getting AI voting advice noted that AI can quickly summarize large amounts of data and provide a potentially unbiased view, can be a timesaver, can customize its responses based on individual priorities, and could explore lesser-known party alternatives. 

ChatGPT’s case AGAINST AI voting advice revolved around AI’s lack of any human understanding, leading it to produce results that lack sociocultural context. It could oversimplify the complexity of political issues, unintentionally reinforce certain perspectives over others, rob voters of their independent critical thinking skills, and, lastly, it could be a privacy concern depending on how the AI’s controllers use your information. 

For question three, ChatGPT provided many of the same reasons to be informed that its counterparts did.  

“An informed voter is more likely to choose candidates and policies that align with their values and priorities. When citizens are knowledgeable about candidates’ positions, policies, and track records, they help ensure that the leaders elected truly represent the will and interests of the population,” the LLM explained.  

It went on to note that informed voters are better at holding their elected representatives accountable and keeping them on track. It said that democratic integrity needs to be protected: 

“Misinformation or ignorance can undermine the democratic process by allowing unqualified or harmful leaders to gain power, which can erode democratic institutions over time.” 

Informed voting, ChatGPT continued, ensures social cohesion stemming from greater engagement in politics, reduces the influence of extremism, fosters personal empowerment, and contributes to the long-term health of a democratic society.