Skip to content

From the archives

Behind the Lines

Two depictions of the Great War

Quite the Catch

Nicholas Herring’s debut

Peggy Who

The long-awaited memoir

For Claude’s Sake!

The computers did not program themselves

Shazia Hafiz Ramji

Humans of AI: Understanding the People Behind the Machines

Joseph Wilson

Aevo UTP

354 pages, hardcover and ebook

Algorithmic Capital: Resistance in the Age of Artificial Intelligence

Jonathan Martineau and Jonathan Durand Folco Translated by David Broder

Verso

496 pages, softcover and ebook

In the mid-1880s, the United States Census Office in Washington commissioned a German American statistician, Herman Hollerith, to develop a system for processing its data. The resulting technology — a punch card tabulating machine — was swiftly adopted, and it replaced hand counting in 1890. The initial design eventually gave way to machine-reading tech like Scantron, which is still used throughout Canada for grading exams. As Forbes put it in 2018, Hollerith and his company, which became known as IBM, “launched a new industry and the fruitful marriage of statistics and computer engineering.” Machine learning — the core of IBM’s business — is foundational for artificial intelligence. While Joseph Wilson doesn’t mention Hollerith, he does name countless others who have followed in his footsteps.

Humans of AI is a surprisingly conversational read that focuses on “the living, breathing humans who make AI work.” In many cases, these are individuals who have been “purposefully erased from the narrative to make it seem like AI is working on its own, a move that boosts share prices and sells products.” An anthropologist in Toronto, Wilson says nothing of the U.S. census and its contribution to modern computing, but that’s not necessarily a flaw of his book: the gap is representative of the myriad and even untraceable ways that AI and its predecessors have been part of our lives.

Wilson’s elegant chapters contribute to the rigorous debate on the distinction between human and artificial intelligence. They are organized around specific roles within the industry: for instance, “The Content Creators,” “The Linguists,” “The Doomers,” and “The Immortalists.” In “The Pioneers,” he revisits the Turing test and the misconception that it was “designed to show whether machines could actually think like humans.” Originally it was meant as a test of “human suggestibility”— how easily we could be duped by a machine. In Wilson’s view, the tragic circumstances of Alan Turing’s arrest in 1952 — for an intimate relationship with a younger man — and subsequent suicide reveal that “measuring intelligence is not the final arbiter of what it means to be fully human; instead, it requires competence in navigating a complex web of social norms.”

An illustration by Karsten Petrat for Shazia Hafiz Ramji’s December 2026 reviews of “Humans of AI,” by Joseph Wilson and “Algorithmic Capital,” by Jonathan Martineau and Jonathan Durand Folco, and translated by David Broder.

Perhaps we’re all getting swept up in the hype.

Karsten Petrat

Examples of AI’s inability to align with cultural conventions appear throughout Wilson’s discussion of algorithmic bias and ethics, a conversation that rightfully “never ends.” One compelling instance was an attempt by Google to show diversity in AI-generated images of people. Even though the “clumsy patch ensured a wide range of skin tones were present,” many of those people “appeared in Nazi uniforms.” Wilson also references a New York Times article about the International Baccalaureate program’s use of AI to predict final grades, instead of conducting exams. Despite being presented as unbiased, the algorithm “tended to lower the existing marks of students from low-income areas, even if they were very high achieving, and raise the marks of students from high-income areas.”

An entire chapter, “The Ghost Workers,” is devoted to labourers in places like Kenya, who are hired by OpenAI to censor graphic content — sometimes for less than $2 (U.S.) an hour. Their assignments can vary from “labeling things in an image” to categorizing online comments. They act as a filter, refining what makes it through to users elsewhere. Wilson interviewed one worker in Nairobi who did “content moderation” within training data sets, which means weeding out erotic, violent, gory, or other inappropriate images. “I don’t know how to explain it,” the man said. “Initially it’s a shock, but after a while it’s just normal work.” Drawing on a 2023 investigation by Time into this “emotionally draining” job, Wilson argues that these moderators have tasks that are potentially as traumatic as those of paramedics or police officers — and should be valued as such.

In 1954, Roald Dahl published “The Great Automatic Grammatizator,” about an engineer who designs a machine that can spit out a prizewinning novel in fifteen minutes. It’s prescient fiction, to put it mildly. In May, Granta magazine published Jamir Nazir’s “The Serpent in the Grove,” which went on to win the 2026 Commonwealth Short Story Prize. Online sleuths argued it had “obvious markers” of AI use — something that Ethan Mollick, a professor at the University of Pennsylvania’s Wharton School, also noted when he ran the text through Pangram. The prize’s previous winner, “Descend,” written by the Montreal author Chanel Sutherland, was also suspect. Sutherland’s Canadian publisher, the House of Anansi Press, defended its author’s originality in the Globe and Mail: “The fact that Pangram is unable to identify something written in a culturally specific voice — in the case of Sutherland’s story, Vincentian oral storytelling tradition — as anything other than AI-generated is further proof of the (well-documented) biases found in AI.”

To unpack the slippage between false positives and actual AI use, Wilson considers books that have been used to train large language models, including ChatGPT. He spoke with the beer sommelier and author Mirella Amato, whose Beerology was found in an OpenAI training data set, along with texts by Margaret Atwood and Rachel Cusk, to name a few. “At the end of the day, I’m not worried I’m going to be replaced by a computer,” Amato said. “My work is multisensory.” Can the same claim be made for literary writing? Anthropic recently announced that its chatbot Claude will watermark AI-generated prose, which could help. But the extent to which it will protect creators remains uncertain.

Looking beyond issues of plagiarism and copyright, Wilson considers contradictory values held by big tech executives. Consider the effective altruism movement founded by William MacAskill, which asks its followers to “interrogate the metrics of any charity they get involved with.” That means an organization receiving money “should be transparent, efficient, and provide proof that the work they do ‘reduces suffering.’ ” Wilson is not so sure, writing, “It seems to me a circular argument though: to know what works, you need to follow the metrics, and to know what metrics to follow, you need to identify what works.” His criticisms of Silicon Valley credos bring levity to such weighty material. Throughout Humans of AI, Wilson centres the people who are involved in the newest form of knowledge production, as well as himself — in the best way possible. His humour and personal anecdotes remind us what truly engaged research and writing look like.

Some of what Wilson covers — from workers in Kenya to effective altruism — also appears in Algorithmic Capital. Likely inspired by Walter Benjamin’s On the Concept of History, Jonathan Durand Folco, who teaches at Saint Paul University in Ottawa, and Jonathan Martineau, a colleague at Concordia University in Montreal, structure their book around twenty theses. They constellate the “strange philosophical galaxy” that “encapsulates several distinct but connected ideas,” such as transhumanism and rationalism that “together form an all-encompassing worldview.” (This tome is hardly aphoristic, as Benjamin’s seminal essay is.) In David Broder’s translation, Folco and Martineau marshal their expansive insights on AI and capitalism into a rigorous “critical theory of algorithms.” In their opening thesis, they write:

It is important to remember that a machine like ChatGPT is only an algorithmic completion tool; it predicts the next word, sentence, or token based on a complex statistical calculation using a huge data set made up of human-generated texts. . . . The ability to analyze a prompt and then generate sophisticated text is surely quite a technical feat — but this does not mean that the machine is “conscious” or “sentient.”

From the outset, Folco and Martineau urge readers to avoid falling for the “mirage of an ethical, responsible AI.” They argue, “As things stand, we can hardly expect AI ethics to deliver convincing results.”

They situate today’s algorithms in the long history of technology, money, and labour — an essential “first step toward a critical theory that avoids technological determinism.” While this declaration sounds measured and academic, their writing is astute and accessible. Taking up the philosopher Nancy Fraser’s concept of an “institutionalized social order,” Folco and Martineau note that our relationships, careers, and purchases all happen within a “vast mesh of apps and data” that normalize surveillance and a loss of control. “Mechanisms for surveillance, prediction, and behavior control” are not a by‑product of hyperconnectivity, they suggest, but “are at the heart of the algorithmic technologies developed by Silicon Valley.”

Algorithmic Capital raises potential local solutions. Heated discussions have developed about new data centres in Alberta and Ontario. Protesters in British Columbia have carried signs with slogans like “Water for the plants, people, and animals, NOT tech-bro billionaires.” In rethinking modern infrastructure, Folco and Martineau discuss how data centres might be integrated within a “digital commons model,” not unlike the telecommunications co‑op in the Quebec municipality of Antoine-Labelle. “The regional county municipality owns the optic fiber infrastructure,” they explain, “and Internet and phone services are provided by a solidarity cooperative composed of member-workers and customers.” It’s a simple, small-scale “public commons partnership” that, they argue, “could be multiplied, or upscaled, at a regional or even national level.” Such concrete visions for resisting algorithmic capital —“to assert the sovereignty of a community over its data”— are energizing.

In one of the later theses on how algorithms are fundamentally shaping our “socialization processes,” Folco and Martineau articulate frank concerns about the role of human behaviour and ego in all of this: “The entire influencer industry consists of turning one’s subjectivity, body, personality, and/or lifestyle into a commodity.” The authors further contribute to contemporary critiques of narcissism with reference to self-promotion on social media: “The desire to be followed, seen, examined, loved, recognized, shared, and commented upon (whether positively or negatively) thus embodies a new form of subjectivity.” Instead of fearing surveillance, we now seek it out for vanity — but also for connection, validation, and income.

Among the many visual elements included in Algorithmic Capital, a simple three-column table stands out. Titled “Algorithmic virtues and vices,” it provides a framework for thinking about ways to mitigate the toxicity of our relationship with our personal devices. Some obvious contenders are listed as virtues, like digital sobriety. But vices are broken out into two categories. “Deficit” is for things like “cyberaddiction” and other extreme forms of intake; “excess” is for isolating measures like “technophobic disconnection.” The table speaks to Folco and Martineau’s focus on promoting awareness and “human flourishing” without unplugging altogether. This idea is central to their final thesis, which encourages a “technosober” way of life that involves “reining in algorithms and tightening the limits of the digital world.”

The book ends with careful, vigorous optimism: “We urge our readers to stop being afraid and to overcome our anxieties, through praxis.” By the conclusion, I was not sure whether I was ready to rave or to sleep, but what is clear is that Folco and Martineau walk their talk. Together, they have exemplified what it looks like to “be technofurious.”

Shazia Hafiz Ramji is the author of the poetry collection Port of Being. She’s writing a novel and dividing her time between Toronto and London.

Advertisement

Advertisement