A document dated 6 November 2025 outlines Library and Archives Canada’s strategic approach to integrating artificial intelligence into its operations. The key messages are:
- LAC embraces AI as transformative for making Canada’s documentary heritage more discoverable and accessible
- Four guiding principles: Transparency, Fairness, Human-in-the-loop, and Trust
- Three strategic pillars: (1) Making collections discoverable through enhanced metadata, (2) Building trust in information by combating misinformation, and (3) Partnering for innovation across sectors
- LAC positions itself as a responsible leader in the GLAM sector, aligning with government-wide digital transformation priorities
The Critique: All Talk, No Show
While establishing a foundational ethical framework is important and necessary, LAC’s document fails to move beyond this preliminary stage. Instead, it reads as if assembled from a corporate buzzword generator—or worse, written by the very AI it claims to be thoughtfully deploying. Here’s why it falls flat:
1. Zero concrete examples: The document mentions “pilot projects” and “concrete actions” but provides not a single specific example. What metadata was enhanced? Which handwriting recognition tool did they test? What were the results? We get platitudes about “accelerating workflows” without evidence that anything has been accelerated.
2. Principles without teeth: The four principles sound admirable until you realize they’re unfalsifiable. “Human-in-the-loop” and “Trust” mean nothing without specifics: Who reviews AI outputs? What happens when bias is detected? What are the actual protocols?
3. Strategic pillars built on air:
- Pillar One promises to “explore” and “pilot” metadata creation—but no timelines, success metrics, or examples of what they’ve actually created
- Pillar Two claims LAC combats misinformation but offers only vague “educational resources” with no description of content or reach
- Pillar Three celebrates partnerships without naming a single AI partner or collaborative achievement
4. Responsibility theatre: The repeated nods to government policies (Digital Ambition, Policy on Service and Digital, etc.) feel like box-ticking rather than meaningful accountability.
5. The irony is deafening: A document about transparency that reveals nothing. A commitment to “showing” how Canadians can engage with heritage that only tells us they’re thinking about it. The lack of self-awareness is almost impressive.
What’s actually missing:
- Specific AI tools or technologies being used or evaluated
- Measurable outcomes and lessons learned from any pilot project
- Budget allocations or resource commitments
- User testimonials or impact data
- Honest discussion of failures or limitations
- Timeline for implementation
This reads less like a position statement and more like a grant application—designed to signal virtue and alignment rather than demonstrate competence or achievement. For an institution tasked with preserving Canada’s documentary heritage, the complete absence of documentation about LAC’s own AI work is remarkably on-brand for government opacity.
And in case you’re wondering, I did use AI to assist in composing this critique.
As a service to LAC, I then asked AI to draft a likely rebuttal. Here it is, unedited.
LAC’s Likely Rebuttal
Re: Concerns Regarding AI Position Statement
We appreciate the thoughtful engagement with our AI position statement and welcome the opportunity to clarify its purpose and scope.
On the Nature of Position Statements
This document was intentionally designed as a strategic framework, not an operational report. Position statements articulate vision, principles, and direction—they establish the “why” and “how” of our approach before diving into the “what.” Detailed project reporting would be premature at this stage and risks creating false expectations about work that is still exploratory and evolving.
On Transparency and Timing
We acknowledge the desire for concrete examples. However, responsible innovation requires careful evaluation before public disclosure. Many of our AI pilots are:
- Still in active testing phases, where outcomes remain uncertain
- Subject to privacy and procurement considerations that limit what we can share publicly
- Being developed in partnership with vendors and institutions under confidentiality agreements
- Dependent on federal approval processes that govern public communication about emerging technologies
Premature disclosure of specific tools, vendors, or methodologies could compromise competitive procurement, create vendor lock-in, or raise unrealistic public expectations.
On Alignment with Government Priorities
The references to federal policies are not “box-ticking”—they represent mandatory compliance frameworks for all federal institutions. The Directive on Automated Decision-Making, for instance, requires algorithmic impact assessments before deployment. Our alignment with these frameworks demonstrates due diligence, not bureaucratic theatre.
On the Scope of This Document
As explicitly stated in the introduction, this position statement does not attempt to address all impacts of AI. Issues including copyright, legal deposit, and definitional questions around creation are being worked through separately with government and GLAM sector colleagues. A comprehensive document covering every dimension of AI would be unwieldy and quickly outdated.
On Substance Over Spectacle
We respectfully disagree that concrete examples are the only measure of serious commitment. The foundational work of:
- Auditing legacy metadata for quality and bias
- Establishing ethical review protocols
- Building staff capacity through training
- Developing partnerships with peer institutions
…is inherently less visible than launching a flashy AI tool, but it is no less essential. Rushing to deploy AI without this groundwork would be the truly irresponsible path.
Looking Forward
We commit to more detailed public reporting as projects mature and move from pilot to implementation phases. Future communications will include case studies, metrics, and lessons learned. However, we will not sacrifice careful planning for the appearance of momentum.
Our position statement is meant to signal readiness and commitment, not premature victory. We invite continued dialogue as our work progresses.
Translation: “We’re a government institution. We write frameworks, not features. Check back in 3-5 years for the progress report that will say we’re still exploring opportunities.”
The above translation paragraph was also composed by AI.