AI Armageddon?
Warnings about artificial intelligence ratcheted up this week, ranging from realistic near-term harms to claims that advanced AI could eventually threaten human survival. Some technology leaders and researchers have described a small but serious possibility that highly capable systems could escape human control or be used to cause catastrophic harm. One widely reported warning suggested a possible 10-per-cent chance of AI causing human extinction within a decade.
How real are the threats?
Already real or highly plausible:
- Deepfakes and political misinformation
- Fraud, impersonation, and cyberattacks
- Unsafe or biased automated decisions
- Job disruption and concentration of wealth
- Privacy loss through large-scale data collection
- AI-generated propaganda and harassment
More uncertain but potentially severe:
- AI systems assisting the development of dangerous weapons
- Autonomous cyberattacks
- Systems behaving in ways their creators cannot predict or control
- A small number of corporations controlling critical information and computing power
Doomsday scenarios are difficult to test because the most advanced systems are private, rapidly changing, and not fully understood. The uncertainty is not proof that catastrophe is likely, but the potential scale of harm explains why some experts argue that governments should act before an accident occurs.
What is Canada proposing?
Canada’s national AI strategy includes more than $2 billion in planned measures for AI capacity, literacy, and adoption. It proposes secure public computing, large Canadian data centres, AI education, and stronger protections for privacy, children and online safety. The government also emphasizes reducing dependence on foreign providers and protecting Canadian technological sovereignty.
Carney has warned that AI could be weaponized against Canadians, but Canada’s earlier Artificial Intelligence and Data Act did not become law, leaving a less complete patchwork of rules at the moment.
Why this matters
In many ways, AI is one of Canada’s technological advantages. Canadian researchers helped create and advance the field, and fostering its continued growth could help address some of the economic weaknesses worsened by the trade war with the US. But AI can also weaken democracy by making false content cheap, convincing, and difficult to trace. If citizens cannot tell whether a speech, photograph, or video is authentic, elections (including the BC and Quebec elections, as well as future referendums on secession) and public debate (around Bill C-39, say) become more vulnerable.
The challenge is to strike the right balance. How can Canada become technologically independent and economically prosperous while also avoiding doomsday scenarios and ensuring AI does not undermine our democracy or sovereignty?
Want to know more?
Better understand the future of work in Careers Chapter Two and the threats posed by AI, social media and surveillance capitalism in Canada at a Crossroads Chapter Five.
Activities by Subject
Social Studies
AI, Democracy & Sovereignty
Reflection questions
- Why can AI-generated misinformation be particularly difficult for a democracy to deal with?
- What happens to public debate if people can no longer easily tell whether an image, video or speech is authentic?
- What does technological sovereignty mean? Why might Canada want to reduce its dependence on foreign AI companies and computing infrastructure?
- Who should be responsible for regulating AI: governments, technology companies, researchers, users (or all of them)?
- How should governments respond when a technology has potentially enormous benefits but also uncertain risks?
- Is preventing a possible future harm a sufficient reason to regulate a technology today? What evidence would you want?
Activity: Who Gets to Control AI?
Review these stakeholders:
- Federal government
- Provincial governments
- Technology companies
- AI researchers
- Schools and teachers
- Businesses
- Indigenous communities
- Workers
- Individual citizens
- Privacy and civil-liberties organizations
Ask:
What should each group be responsible for? What decisions should each group not be able to make alone? What could happen if one group had almost all the power? Where should the balance between innovation, safety and individual freedom sit?
If AI becomes essential infrastructure, who should have the power to control it?
Science
Risk, Uncertainty & Systems
Reflection questions
- What is the difference between a threat that is already occurring and a hypothetical future threat?
- Why are some AI risks easier to study than others?
- How could scientists test whether an AI system is behaving unpredictably?
- What kinds of evidence would make you more concerned about an AI risk?
- Should the potential scale of a risk affect how much research we do to understand it?
Activity: The AI Risk Radar
Review this list of risks:
- Deepfakes
- Fraud
- Privacy loss
- Job disruption
- Biased decisions
- AI-assisted cyberattacks
- AI-assisted weapons development
- Autonomous systems
- Loss of human control
- Concentration of computing power
Place each on a four-part grid:
Already happening → Plausible → Highly uncertain → Unknown
Then add a second dimension:
Low potential impact → High potential impact
Then ask:
Should something be investigated because it is likely, or because the consequences could be enormous?
English
Truth, Persuasion & AI-Generated Information
Reflection questions
- What makes a piece of misinformation convincing?
- How does the phrase ‘AI Armageddon‘ affect the reader before they even encounter the evidence?
- What is the difference between warning about a risk and predicting an outcome?
- How can headlines exaggerate—or minimize—the same information?
- What happens to storytelling when the audience cannot be certain who created the content?
- How should journalists identify AI-generated or AI-assisted material?
Activity: One Story, Three Headlines
Read this basic statement:
Researchers and technology leaders are warning that increasingly capable AI systems could create serious risks, while governments are developing policies to manage those risks.
Write three headlines:
1. Alarmist
Designed to make readers worried.
2. Optimistic
Designed to emphasize opportunity.
3. Neutral
Designed to inform without telling the reader how to feel.
Then compare:
- What words changed?
- What information was emphasized?
- What information disappeared?
- Which headline most clearly separates evidence from interpretation?
Bonus: Can You Tell It’s Fake?
View several fictional AI-generated posts, images or headlines and ask:
What evidence would you need before sharing this?
Create a classroom verification checklist:
Source → Evidence → Date → Original context → Other sources → Who benefits? → What is missing?
Visual Storytelling/Art
When Reality Can Be Manufactured
Reflection questions
- How has photography traditionally been associated with evidence or truth?
- What happens to that assumption when photographs can be generated?
- Can an AI-generated image still be art?
- Who is the artist when a person gives instructions to an AI system?
- How could artists use AI without allowing AI to replace human creativity?
- Can an artwork communicate something important even when the image isn’t literally real?
- If almost anything can be generated, what does an image actually prove?
Activity: Real, Fake or Impossible?
Create a series of fictional images representing:
- A real Canadian landscape
- An altered landscape
- An entirely AI-generated landscape
- A future landscape that doesn’t exist yet
Don’t simply guess which is which.
Instead, ask:
What clues did you use?
Then ask:
What happens when visual evidence is no longer automatically trustworthy?
Creative challenge: Design the AI Future
Create one image representing a possible future of AI.
But you must include both an opportunity and a risk.
For example:
- AI improving healthcare + privacy concerns
- AI helping workers + job displacement
- AI improving education + misinformation
- AI helping scientific research + loss of human oversight
Careers
Working With AI, Not Just Competing With It
Reflection questions
- Which parts of a job are easiest for AI to automate?
- Which parts still require human judgment, relationships, creativity or physical presence?
- Could AI eliminate some jobs while creating entirely new ones?
- Who benefits economically from increased AI productivity?
- What skills might become more valuable as AI becomes more capable?
- How should schools prepare students for a labour market that may change rapidly?
Activity: The AI Job Audit
Choose a career.
Break the job into 10 tasks.
For each task, classify it:
- AI could probably assist
- AI could assist but needs human oversight
- Human judgment/interaction is essential
- We don’t know yet
Then ask:
- What would the job look like if AI handled the green tasks?
- What new skills would the worker need?
- What could go wrong if AI made decisions without human oversight?
- Would the job disappear or change?
Cross Curricular Activity
The AI Decision
Imagine Canada is deciding how quickly to expand AI development and adoption. You are advising the government. You cannot simply say ‘AI is good’ or ‘AI is bad.’ You need to identify what Canada needs to know before making decisions.
Each course gets one responsibility:
Science: What are the actual risks? What evidence exists? What remains uncertain?
Social Studies: Who should make the rules? Who gets power, and how should citizens be protected?
English: How do we know what’s true? How can Canadians distinguish evidence from misinformation and persuasion?
Art: What happens to creativity and visual truth? How does generative AI change the meaning of images, art and authorship?
Careers: What happens to work? Which jobs and skills could change, and who benefits?
Then bring the five groups together.
Each group must identify:
1 thing we know
1 thing we don’t know
1 risk
1 opportunity
1 question Canada should answer before acting