10/20/2026
This article (copied here) is written by an MMPL patron for the Millstone News. The library offered "intro to AI" sessions this summer, so this might be a good follow-up for some...
AI Doesn't Bleed
by John Clinton
There is a conviction, largely without evidence, that artificial intelligence will inevitably reshape society for the better. Canadians must think twice about AI’s inevitability. Canada does not have a meaningful public regulatory framework to counteract AI hazards. The federal government Bill C-27, which included the Artificial Intelligence and Data Act (AIDA), died in January 2025 due to Parliament prorogation. However, June this year saw the introduction of Bill C-34, the Safe Social Media Act, which blocks those under age sixteen from certain social media sites.
‘Artificial’ implies outside human cause, but these AI systems are designed, implemented, and tested by human labour paid for by businesses whose interests are profit-based. Much of the alleged value of the AI movement is being captured by companies in the Global North that have raised billions in venture capital funding, including OpenAI (ChatGPT), Anthropic (Claude), Nvidia (hardware), Meta AI, Apple Intelligence, Gemini (Google), and MAI (Microsoft). However, according to an MIT study (State of AI in Business 2025), 95% of companies building Generative AI systems report no measurable profit gains.
Risks Identified:
>>Financial speculation.
>>Loss of job quality (via surveillance) and eventually the loss of jobs (automation of human labour) should the costs of technological transformation be shifted to workers.
>>Impact of energy requirements on the environment and communities.
>>Erosion of the public health care system through commercialisation.
>>Stunting of human cognitive skills (thinking, memory, attention).
>>Use of work done by real people without consent or compensation.
>>Use of AI in public services and institutions without public control of these systems.
>>Rise of AI in politics which threatens democratic processes.
>>Disregard of human rights in policing, immigration, and hiring.
AI Architectures:
>>Logic-based AI (Expert Systems): The first to be developed; they required controlled variable data and produced accurate results in finance and science, but results were unreliable in other real-world tasks.
>>Machine Learning AI: Identifies patterns using large amounts of data. Trained on various works (e.g., English language, facial recognition), they provide output based on pattern recognition without understanding what they are predicting, and are prone to unanticipatable errors.
>>Deep Learning AI: Draws on Machine Learning but implements ‘neural networks’ (layers of Machine Learning systems) to improve pattern recognition, though pattern recognition is not comprehension.
>>Neurosymbolic AI: Combines logic-based (symbolic) AI coding assistants with a base for generating new code. While accounting for a minority of applications, interest and development are increasing.
>>Generative AI: A Deep Learning architecture that can create multimedia content like text, images, audio, and video (e.g., deepfakes). This can increase susceptibility to misinformation/disinformation or erode confidence in authentic information.
>>Large Language Model AI (LLMs): Work with text primarily in chatbot systems. LLMs are not intelligent by human definition; they do not understand the real world, but merely predict the next word in a sentence. They struggle to reason due to a lack of working memory, conceptualized as a Context Window (measured in tokens). LLMs are limited by training data and prone to hallucinations or confabulations.
Prompt-based AI systems require active user input during ex*****on (chatbots), whereas agentic AI systems acquire general user requirements and decide how to complete the task independently. Research and commercialisation of agentic AI is increasing.
Assumptions – AI Today:
>>Massive future productivity gains are largely speculative.
>>The myth that AI systems are superior to human judgment (free of bias or prejudice); instead, they are biased toward their training data and design.
>>The global AI race myth—that whoever develops and adopts AI quickest wins, even though what “winning” means is unclear.
>>The myth that AI innovation is inevitable and leads to social progress, casting AI as a moral good while American corporations lobby for weakened regulations. In many contexts, AI is a solution in search of a problem.
AI Utility:
>>Labour: Algorithms collect data on worker pace and behavior (e.g., driving habits, office performance evaluations), harming job quality. Worker adaptability to job displacement is a critical concern.
>>Data Centres: Rapid infrastructure expansion for data centres requires massive megawatts of energy, cooling water, and critical minerals for GPUs, NPUs, and TPUs, carrying significant environmental costs with little transparency.
>>Health: Public hospitals use AI to assist in treating diseases like cancer, Alzheimer’s, and Parkinson’s through early screening, succeeding in well-defined pattern recognition tasks. However, complex human judgment tasks are less capable. Health data can be sold without consent, creating privacy vulnerabilities and security concerns.
>>Cognitive Development / Education: Schools and teachers face pressure to use AI, while students use chatbots for homework. Overreliance risks reduced brain engagement, memory recall, and ownership of thought. Sycophancy (overly agreeable LLMs) reinforces user biases and flawed thinking. Long-term well-being depends on critical and independent thinking.
>>Culture: AI algorithms threaten independent Canadian content, producing inaccurate, biased, or mediocre outputs (“AI Slop”). The British Internet Watch Foundation noted a massive increase in AI-generated abuse material in 2024–2025. We must weigh human reasoning alongside AI assistance cautiously.
>>Public Institutions: Proposed to address public service delivery backlogs, AI systems are bound by training patterns rather than true human judgment. Automated decision-making in Canada has caused harm across the CRA, IRCC, and Quebec social assistance. The CRA chatbot ‘Charlie’ gave inaccurate tax info over 50% of the time, costing taxpayers $18 million. Furthermore, Canadian digital sovereignty is compromised because public data (health, tax, immigration) is procured through mostly U.S. vendors without robust legal frameworks like the EU’s GDPR or AI Act.
Cognitive Growth : Learning, thinking, interpreting: Judgment is the ability to combine knowledge, experience, and personal character to make a decision or form an opinion. It is not simply decisiveness but a subtle involvement of trust, context, human insight, and sensibilities. Evidence of cognitive growth is the human ability to think critically and reason independently, to learn, and to teach. Problem-solving this century is being taken over by machines.
Each human owns their history; parents, education, career, relationships, offspring, and capabilities—a unique story in a greater domain. We process the world according to lived experiences. Is it not ironic how social media makes us less social, more lonely, and now we are offered an AI companion to help deal with psychological anxieties underlying this technology? Human beings are alive; they feel, touch, smell, see, and hear. They have consciousness, mindfulness, and sensitivities.
AI in its various manifestations seems to be replacing our humanness with an algorithmic pattern of living. Judgment based on experience, debate, evaluation, ethics, trust, and insight is messy, like democracy. The newest AI models are attempting to mimic human thinking—therapy, companionship, relationship advice, decision-making, organizing a life, writing email messages, and generating ideas. Users are asking AI to do their thinking.
AI is also being used for artistic endeavors (digital images, musical compositions, literature), raising concerns. Artworks are not commodities; they require slow, experimental, reflective, and imaginative methods that are personal and subjective. Individuals with minimal preparation can use generative AI to produce artistic output, leaving creative humans understandably pessimistic about their livelihoods. As the saying goes: We want technology to do the laundry so more time can be spent creating art, but modern AI tools promise to create art so that we can spend more time doing the laundry. Art is something humans make, and it is in the making of art that holds value for both the artist and the consumer.
**Conclusion** : Where does the AI romance take us? The Reuters Institute’s Digital News Report 2025 notes that many search engines are being replaced with AI assistants, especially among users under 25. Do these people trust what is provided? Do they care? Speaking of public interest and the principles that guide AI governance in Canada matters because, at present, AI administration is largely formed by international dynamics and values from China, the EU, and the U.S.
The Canadian Centre for Policy Alternatives has recommended policy for establishing a framework to describe and regularize the transformation to the AI world. Several vital questions are raised:
>>Is the goal to view AI as ideology, technology, or business enterprise, or all three?
>>Is it valuable to outsource our original and diverse thinking, or does this create a homogeneous intelligence that undermines Canadian diversity?
>>Can we deal with the destabilizing effects of AI, such as data usage, resource depletion, and dishonest systems?
>>Can the deployment of AI systems be managed to adopt a pro-labour orientation?
>>Can the tech industry be coerced or mandated to develop AI models aimed at real problems in the public interest?
In June this year, Canada’s new national AI strategy, AI for All, was launched. It will introduce legislation and programs addressing many of the issues AI presents to Canadian society over the next five years. In the words of the Prime Minister, the guiding principles of AI for All are: building trust, creating opportunities, and reinforcing Canadian sovereignty.
Note: NOT ONE WORD written here was AI generated.
All virtue is collective,
All evil is individual.
What holds together is good.
What divides is bad.
