Briefing Notes – Ethics of AI Institute Meeting

Date: October 02, 2026
Time: 3:00 PM EST
Chair: Matthew Silk
Attendees: Members of the Ethics of AI Institute

White Paper Initiative

• The Institute is launching a new white paper series to convert meeting discussions into publicly accessible publications.
• Recruitment is underway for Editor-in-Chief and Communications Officer positions.
• Members were encouraged to volunteer as writers or contributors.

Potential AI Literacy Symposium

• The Institute is exploring hosting a symposium in Winter 2027, potentially focused on AI literacy.
• Interest was expressed in involving participants from both the Mount Allison and Waterloo groups.
• We will investigate grant opportunities to support the event.

General Thoughts on the Question

While members acknowledged that AI poses significant risks, there was broad skepticism toward claims that AI is likely to independently cause human extinction in the near future.

Many participants distinguished between:

• AI as an autonomous existential threat ("rogue AI")
• Humans using AI in dangerous ways

The latter was generally viewed as the more realistic concern.

Concerns About AI Development Speed

• AI development may be advancing faster than society's ability to understand and regulate it.
• Lack of adequate oversight.
• Insufficient public discussion of long-term implications.
• Deployment of increasingly powerful systems before safety structures are mature.

Skepticism Toward Existential-Risk Claims

• Predictions such as a "10% chance AI will wipe out humanity" are speculative.
• Humans are poor forecasters of long-term technological outcomes.
• Such statements may function as public relations or marketing tools.
• Existential-risk narratives: increase public attention, may encourage regulation; could exaggerate AI capabilities and create unnecessary fear.

Corporate Incentives and "Cry Wolf" Concerns

Distrust of AI companies making catastrophic predictions about their own products.

• Companies may benefit by portraying AI as extraordinarily powerful.
• Calls for regulation could allow major firms to shape regulatory frameworks in their own interests.
• "Rogue AI" narratives risk providing corporations with plausible deniability if harms occur.
• Analogy was made to: climate change regulation; tobacco industry messaging; "Too big to fail" financial institutions.

Transparency and Explainability

There’s promising research aimed at opening AI "black boxes" through:

• Explainability techniques
• Tracking decision pathways
• Reverse-engineering neural network outputs

Benefits identified:

• Improved accountability
• Better implementation of AI regulation
• Stronger legal oversight

Robotics as a Distinct Risk

AI software alone is less concerning than AI combined with physical systems.

• Risks increase substantially when AI gains physical capabilities.
• Preventing inappropriate integration with robotics may be more important than limiting AI models themselves.

Concerns included:

• Autonomous weapons
• Industrial robotics
• Military applications
• AI controlling critical infrastructure

Human and Social Harms

There are several near-term concerns that may be more significant than extinction scenarios:

Employment and Identity

• AI may replace specialized professional work.
• Workers may experience loss of purpose or identity.
• Students entering certain fields may face uncertainty about career prospects.

Human Relationships

• Increased reliance on AI assistants could reduce meaningful human interaction.
• Younger generations may become accustomed to automated rather than human support.
• Participants worried about erosion of social and interpersonal skills.

Information Integrity

Concerns were raised about:

• Manipulation of information
• Historical revisionism
• Control of online knowledge repositories
• Ownership of cultural and intellectual works

Critique of "Artificial Superintelligence"

The idea of superintelligence received substantial scrutiny.

• Definitions are often vague.
• Comparisons between human and machine intelligence may be misleading.
• AI may exceed humans in narrow capabilities while lacking empathy, social understanding, physical-world knowledge, moral reasoning, relational capabilities.
• AI intelligence and human intelligence may be fundamentally different and not directly comparable.

Problems with AI Language and Anthropomorphism

Participants expressed concern about language such as: “AI… thinks, wants, is determined, collaborates, deceives, blackmails, goes rogue”.

Arguments raised:

• Such language encourages anthropomorphism.
• Media coverage often frames AI as human-like.
• This can distort public understanding of actual AI capabilities.

Suggestions included:

• Using more technical language.
• Referring to AI as statistical or predictive systems.
• Emphasizing capability rather than intention.

Accountability and Governance

Discussion of who should be responsible when AI systems cause harm: developers? corporations? operators? regulators?

A common view emerged that:

• Accountability should remain with people and institutions.
• AI should not become a mechanism for diffusing responsibility.

Suggestions included:

• Independent oversight bodies
• External governance mechanisms
• Regulatory frameworks separate from AI companies themselves

Ownership of AI and Human Knowledge

A broader ethical discussion examined:

• Who benefits from AI systems trained on humanity's collective knowledge.
• Whether a small number of corporations should control AI infrastructure.
• Whether society should receive broader benefits from AI-generated value.

Participants questioned:

• Ownership of training data
• Concentration of economic gains
• Distribution of benefits from AI systems

Future Meeting Topics

Should AI be nationalized?

• Public versus private ownership.
• Government stewardship of AI resources.

AI and Scientific Research

• Use of AI in scientific discovery.
• Explainability requirements in science.
• Intellectual property and research ethics.

Surveillance and AI Governance

• Regulatory structures.
• Accountability mechanisms.
• Public oversight.

Action Items

• Follow up on symposium grant opportunity.
• Continue recruiting editor and communications officer positions.
• Explore development of a white paper based on the existential-risk discussion.
• Schedule next meeting for November 6, 2026 (tentative).