blair
LightNews LightNews
banner
blairaf.com
blair
@blairaf.com
3.3K followers 260 following 910 posts
Asst. Prof, Amii Fellow, sci-fi creator @ UAlberta

Transfeminist AI governance, ethics, policy: blairaf.com

Gay trans cyborg warriors vs. AI monsters: objecttype3.app
Posts Replies Media Videos
Pinned
blairaf.com
blair @blairaf.com · Apr 10
Why is AI governance so often ineffective at preventing AI from harming people and the planet? My new paper Transfeminist AI Governance addresses this question, now out in this month's issue of First Monday: firstmonday.org/ojs/index.ph...
Abstract for a paper entitled "Transfeminist AI Governance". The abstract reads: "This paper re-imagines the governance of artificial intelligence (AI) through a transfeminist lens, focusing on challenges of power, participation, and injustice, and on opportunities for advancing equity, community-based resistance, and transformative change. AI governance is a field of research and practice seeking to maximize benefits and minimize harms caused by AI systems. Unfortunately, AI governance practices are frequently ineffective at preventing AI systems from harming people and the environment, with historically marginalized groups such as trans people being particularly vulnerable to harm. Building upon trans and feminist theories of ethics, I introduce an approach to transfeminist AI governance. Applying a transfeminist lens in combination with a critical self-reflexivity methodology, I retroactively reinterpret findings from three empirical studies of AI governance practices in Canada and globally. In three reflections on my findings, I show that large-scale AI governance systems structurally prioritize the needs of industry over marginalized communities. As a result, AI governance is limited by power imbalances and exclusionary norms. My reflections reveal that re-grounding AI governance in transfeminist ethical principles can support AI governance researchers, practitioners, and organizers in addressing those limitations."
Reposted by blair
blairaf.com
My new article on Canada's recent AI strategy consultation. Lightning-fast engagement, dubious data protection measures, black box analysis, & AI-generated summaries from made-in-USA LLMs are not a winning strategy for building public trust

betakit.com/canadas-new-...
Canada’s new AI strategy is off to a bad start | BetaKit
Canadians already have low trust in AI. Exclusionary and unclear public engagement methods aren’t helping.
betakit.com
February 9, 2026 at 9:32 PM
Reposted by blair
golikehellmachine.com
even if i had absolutely no knowledge of elon musk or his history, i would have a lot of questions about someone who claimed they were going to build settlements on the moon within a decade, like, “what experience do you have in doing this on earth?”
February 9, 2026 at 9:57 PM
blairaf.com
My new article on Canada's recent AI strategy consultation. Lightning-fast engagement, dubious data protection measures, black box analysis, & AI-generated summaries from made-in-USA LLMs are not a winning strategy for building public trust

betakit.com/canadas-new-...
Canada’s new AI strategy is off to a bad start | BetaKit
Canadians already have low trust in AI. Exclusionary and unclear public engagement methods aren’t helping.
betakit.com
February 9, 2026 at 9:32 PM
Reposted by blair
alexhanna.bsky.social
The main conceit of "AI" (namely chatbots and "agents") are the twin promises of control and productivity.

The mechanism of control is that you can be replaced more effectively by a machine, or someone who wields the machine more adroitly than you.
volts.wtf David Roberts @volts.wtf · 2d
The bosses are forcing their employees to use it. Politicians are competing to shovel money & infrastructure at it. The billionaires who control media are constantly lecturing & hectoring people to accept it & use it more.

They don't care it's losing money. They don't care people hate it.
February 8, 2026 at 12:15 AM
Reposted by blair
internethippo.bsky.social
I think "woke" was at least partly identified with "irritating" and the way we're rectifying that in Woke 2 is by being scary instead
February 6, 2026 at 2:52 AM
Reposted by blair
sayambular.bsky.social
I used to love computer it was my friend. Now I have hate in my heart
February 5, 2026 at 9:17 PM
blairaf.com
Reading week plans
Simpsons screenshot of book titled So You've Decided to Become Isolated and Weird
February 7, 2026 at 6:48 PM
Reposted by blair
wandrew.bsky.social
'Should we go all in on AI? Let's ask AI.'
February 7, 2026 at 3:30 AM
Reposted by blair
blairaf.com
Me in 2021: I wish we had a more integrative national AI strategy and more public consultation on the AI strategy

Monkey paw in 2026:
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 6, 2026 at 4:43 PM
blairaf.com
Spending Friday night eating fried chicken and writing an angry essay about the government life is great
February 7, 2026 at 12:54 AM
Reposted by blair
orientaljanedoe.bsky.social
ICYMI:

950+ Google workers signed a petition demanding that executives "disclose all contracts and collaboration with DHS, CBP, and ICE, and to divest from these partnerships."
www.nytimes.com/2026/02/06/b...
Google Workers Demand End to Cloud Services for Immigration Agencies
www.nytimes.com
February 6, 2026 at 11:37 PM
Reposted by blair
trans.bsky.social
art has already been democratized by innovations like digital drawing tablets , libre/ FOSS image editing programs, & the internet.

we have incredible creative tools available for either free or very inexpensively. democracy requires effort & work!
February 6, 2026 at 9:51 PM
Reposted by blair
sebthecanadian.ca
This is without a doubt one of the best things I’ve read in a while.
blairaf.com blair @blairaf.com · Apr 10
Why is AI governance so often ineffective at preventing AI from harming people and the planet? My new paper Transfeminist AI Governance addresses this question, now out in this month's issue of First Monday: firstmonday.org/ojs/index.ph...
Abstract for a paper entitled "Transfeminist AI Governance". The abstract reads: "This paper re-imagines the governance of artificial intelligence (AI) through a transfeminist lens, focusing on challenges of power, participation, and injustice, and on opportunities for advancing equity, community-based resistance, and transformative change. AI governance is a field of research and practice seeking to maximize benefits and minimize harms caused by AI systems. Unfortunately, AI governance practices are frequently ineffective at preventing AI systems from harming people and the environment, with historically marginalized groups such as trans people being particularly vulnerable to harm. Building upon trans and feminist theories of ethics, I introduce an approach to transfeminist AI governance. Applying a transfeminist lens in combination with a critical self-reflexivity methodology, I retroactively reinterpret findings from three empirical studies of AI governance practices in Canada and globally. In three reflections on my findings, I show that large-scale AI governance systems structurally prioritize the needs of industry over marginalized communities. As a result, AI governance is limited by power imbalances and exclusionary norms. My reflections reveal that re-grounding AI governance in transfeminist ethical principles can support AI governance researchers, practitioners, and organizers in addressing those limitations."
February 6, 2026 at 9:01 PM
blairaf.com
Me in 2021: I wish we had a more integrative national AI strategy and more public consultation on the AI strategy

Monkey paw in 2026:
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 6, 2026 at 4:43 PM
Reposted by blair
blairaf.com
Another thought: did any part of the consultation form obtain participant consent for feeding their responses into an opaque daisychain of models developed by foreign tech companies? Was PII filtered out of all 64k responses before pushing them through the pipeline?
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 3:40 AM
Reposted by blair
blairaf.com
Huge missed opportunity for Canada to show global leadership in AI governance. We could be a frontrunner on so many participatory methods - open legislative development and policy workshops, public awareness campaigns, citizens assemblies - instead we're making AI slop and calling it innovation
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 1:36 AM
blairaf.com
Another thought: did any part of the consultation form obtain participant consent for feeding their responses into an opaque daisychain of models developed by foreign tech companies? Was PII filtered out of all 64k responses before pushing them through the pipeline?
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 3:40 AM
Reposted by blair
dryad.technology
They asked for a consultation on AI and then replaced the results of the consultation with an AI output and from this it looks as if my input that they should not use AI was somehow not part of the AI output
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 3:34 AM
Reposted by blair
stanleyparks.bsky.social
The lack of real introspection about what kinds of data are actually inside a lot of these models, and the failure to develop proper auditing frameworks is really problematic. Without these, government is just funding hype, and that is not any kind of sustainable investment.
February 5, 2026 at 3:29 AM
Reposted by blair
peterlouwe.bsky.social
I did the gov't questionnaire and every question was loaded with the assumption AI would be used in every ministry. There was no room to consider IF AI was appropriate.
February 5, 2026 at 3:28 AM
Reposted by blair
blairaf.com
If your goal as a government is to build public trust, throwing public feedback into an opaque analytics pipeline seems counterproductive to that goal
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 4, 2026 at 8:59 PM
Reposted by blair
aroddick.bsky.social
Wild. It is EXACTLY like university administrators approach.
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 3:10 AM
Reposted by blair
parismarx.com
canada continuing to drop the ball on taking tech policy seriously
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 3:02 AM
blairaf.com
Inverse transfem voice training to make my voice lower and dykier
February 5, 2026 at 2:45 AM
blairaf.com
Huge missed opportunity for Canada to show global leadership in AI governance. We could be a frontrunner on so many participatory methods - open legislative development and policy workshops, public awareness campaigns, citizens assemblies - instead we're making AI slop and calling it innovation
blairaf.com blair @blairaf.com · 5d
Canada did a consultation on a new national AI strategy, formed an expert task force to write 32 reports, then used AI to analyze the responses & reports. The result is a summary that strings together 100s of vague action items & flattens nuance and policy trade-offs into false consensus
Screenshot from Government of Canada website that reads: "Over 11,300 participants provided their thoughts and ideas during the 30-day online public consultation, which ran from October 1 to 31, 2025. We captured over 64,600 responses to 26 questions on the eight AI pillars identified for the purpose of the consultation. The responses were analyzed using SimpleSurvey, which uses natural language processing algorithms to provide insights from open-ended questions by analyzing sentiments, keywords and topics. ISED also created an internal classification pipeline for analysis, incorporating Cohere Command A, OpenAI GPT-5 nano, Anthropic Claude Haiku and Google Gemini Flash to read through the submissions and identify common themes. Responses were also classified by question, pillar and sentiment. Human reviewers validated and refined the AI-generated analyses, ensuring accuracy and comprehensive representation of all perspectives. The same approach was used to analyze the Task Force reports. This innovative approach enabled thorough, unbiased reporting in record time—months faster than traditional methods—demonstrating a balance between efficient delivery and responsible, Canadian-centred AI adoption."
February 5, 2026 at 1:36 AM