Part one, See, opens in a Vermont cotton mill in August 1910.
Part two:
On 21 May 1930, at the Dharasana Salt Works on the Gujarat coast, United Press reporter Webb Miller saw men walk up to a line of police and get beaten.
The volunteers moved in lines toward the barbed-wire fences around the salt pans. Around 400 police waited for them. When the volunteers would not leave, the police attacked with clubs. Miller wrote in his report that “the volunteers made no resistance“ and took the blows “with almost unbelievable meekness.” Sarojini Naidu, who spoke to them before they began, had told them: “You must not use any violence under any circumstances.“ The report was sent from DHARASANA and said 260 people were injured.
Gandhi was not present. Police had arrested him on 4 May, so he was in prison while others continued the campaign.
A part of the story is often forgotten. Reports say British telegraph operators delayed Miller’s first attempts to send his report. The story was sent only after Miller threatened to reveal the censorship. It appeared in over a thousand newspapers, and Senator John J. Blaine even read it into the United States Senate record.
The violence at Dharasana would have happened even without Miller. What made the difference was having a witness, sending a cable, and insisting the story get out.
That is the hinge between event and public judgment.
So why does a salt works matter in a discussion about artificial intelligence?
The first part of this series, See, started with a photo of a girl in a Vermont cotton mill and asked whose decisions affect. This part moves to the next two steps in the method Joseph Cardijn created, and Pope John XXIII included in Mater et Magistra on 15 May 1961: first, form a judgment, meaning judge in the light of the same principles; then act on what can be done.
Dharasana shows both steps at once. The campaign realized that injustice depends on consent, so they refused to give it. Then they acted in a way the world could see. Because someone shared their story beyond the salt pans, the volunteers’ actions became a message.
Marshall McLuhan spent his career exploring this idea. He said a medium is not just a neutral channel. Its “personal and social consequences“ change how people live and interact. A telegraph office is a medium, and so is an AI assistant that filters, ranks, and answers for us. Here, medium means the channel that carries action and meaning, including how AI shapes judgment and action.
Judge: ask what the medium does
McLuhan gave us a way to judge technology, which he and his son Eric described in Laws of Media: The New Science (University of Toronto Press, 1988). Ask four questions about any medium or AI system: What does it enhance? What does it make obsolete? What does it bring back from the past? What does it turn into when pushed to the extreme? Use these questions to judge the medium and the AI system itself.
Here is an initial draft about an AI assistant, meant for group discussion:
Enhances
Recall, drafting, search, and pattern-finding at a speed no person matches.
Obsolesces
The entry-level task where people once learned a trade by doing the dull part.
Retrieves
The oracle and the scribe: someone who answers every question and writes down what you say.
Flips into
A record that remembers for you what you ought to remember, and a recommender that tells you what people like you did next.
Merton adds another test, the third of his four steps for reading any tradition: honest distinction. This week’s Hindu material and Merton’s Gandhi give us the words. Dharma means “the right thing to do here, for someone like me, in this situation.” An AI recommendation engine tells you what people like you did in similar situations. These may sound similar, but they are not the same. Judging means explaining where they differ.
Catholic social teaching supplies the yardsticks: the dignity of the person, the principle of subsidiarity (decisions made at the most local level capable of making them), and solidarity. Apply subsidiarity to the rental market. RealPage’s software suggested rents to landlords in many local markets. New York banned such rent-setting algorithms, effective December 2025. Then, on 29 September 2026, Multifamily Dive reported that a federal judge paused the ban, treating the software’s price recommendations as commercial speech. The pause is temporary while the court weighs RealPage’s First Amendment claim, showing how AI recommendations can shape rents.
When a company uses AI to decide a community’s rents and residents have no say, the community loses part of what gram swaraj, or village self-rule, is supposed to protect. Judgment means pointing out that loss and explaining why it is important. Here, gram swaraj means local self-rule and shared control.
Judge the ground, not only the figure.
In part one, figure and ground referred to the screen and the technology behind it. The background deserves its own judgment. Figure means what is immediately visible; ground means the context behind it.
Synergy Research Group reported that global cloud infrastructure revenue was $143.4 billion in the second quarter of 2026. Other sources say Amazon had 28 percent, Microsoft 20 percent, and Google 15 percent, for a total of about 63 percent. From there, Swadeshi, Gandhi’s idea of supporting local control and resisting concentrated power, raises the question: What does it mean for a school, parish, or small country to choose local options when three companies take 63 cents of every cloud dollar?
Seva, or service, needs people, places, and teachers. The Futurum Group estimated in February 2026 that five companies plan to spend about $660 to $690 billion on infrastructure in 2026. In contrast, the Organization for Economic Co-operation and Development (OECD) reported in April 2026 that official development aid from donor countries dropped 23.1 percent in 2025, the largest annual decrease ever recorded.
These two numbers do not show cause and effect. Still, together, they raise an important question: What happens when most resources go into just one type of system?
Act: Find the lever, and ask who is left out.
Noncooperation is the act that the Week 5 slides in my class highlight. It is Gandhi’s withdrawal from unjust systems, which the slides call “a structural critique that algorithms cannot perform,” because “moral agency requires a person, not a process.” That is also why the AI question matters here. Here, noncooperation means refusing unjust systems in a way that makes a moral point, including when the system is an AI system.
The Dharasana withdrawal worked because it had both a witness and a way to share the story. Hollywood writers used a similar approach in 2023. For 148 days starting 2 May, they went on strike and won a contract: AI is not a person, its output is not literature, and no writer has to use it. Whether AI can use their work for training remains undecided, so the dispute still centers on how AI affects writers’ work.
Writers had power because studios needed their work. Most people do not have that kind of leverage. The outplacement firm Challenger, Gray & Christmas counted 112,713 US job cuts through July 2026 that mentioned AI, about 24 percent of 477,033 total cuts. Employers’ reasons aren’t checked, and the numbers don’t show who can push back.
Merton believed Gandhi taught that political action is about witnessing to the truth, not just winning. As the slides say, “success was secondary to fidelity.” This shifts the focus: it is not just about winning, but about staying true to the people involved and making sure their story is heard. Here, fidelity means staying true to the person affected.
McLuhan’s warning fits here: “We look at the present through a rearview mirror. We march backward into the future.“ Institutions often use old tools like boycotts, contract clauses, or laws made for different times because they know them. These actions work only when the tool matches the new situation, which is less often than we hope.
Merton, writing about technology, put the stakes in a sentence: “It does us no good to make fantastic progress if we do not know how to live with it, if we cannot make good use of it, and if, in fact, our technology becomes nothing more than an expensive and complicated way of cultural disintegration.“
The strongest argument against saying no
The strongest objection comes from the other side of the table, and it comes in two parts.
Supporters say data center funding comes from private shareholders and was never meant for hospitals, and that donor governments chose to cut aid, not data centers. They also argue that these machines could lower the cost of diagnosis, tutoring, and logistics for groups that help people experiencing poverty, so the buildings might help with some of the work that seva has done alone. The real question is what gets built and who decides. No one has proven that one type of spending caused the other, but the AI stakes are plain. Here, the point is to distinguish spending decisions from spending effects.
Gandhi’s boycott of British cloth put mill workers out of work, even though he didn’t want to influence them. Lancashire is an example. So refusing AI systems completely could let the builders control everything, leaving workers to bear the consequences with no say. The objection is that a boycott works only for people with choices, and the cost often falls on the wrong people.
This objection is partly correct, and that part is important. If a refusal ignores who pays the price, it repeats the harm it tries to stop. That is why seeing comes first. Even so, we do not have to accept everything. Dharasana and the writers’ strike both show people saying no while also naming who is affected.
Four steps for a Monday
Here is a version short enough to fit on an index card:
Go through the four questions. Ask what the technology improves, what it replaces, what it brings back from the past, and what it turns into when pushed to the limit. Write one sentence for each answer. Use these questions to judge the medium clearly.
Name the principle. Say whether dignity, subsidiarity, or solidarity is at stake, and state the reason aloud. Be clear about which principle applies.
Find the lever and the missing person—list who can withhold something the system needs, and who cannot. Use these terms to identify who has power and who does not.
Measure fidelity. Decide what it means to stay true to the person affected, and consider that along with the usual numbers. Keep fidelity tied to the person.
The story of Dharasana almost disappeared at a telegraph office window.
Think about who owned that window. The truth about the violence had to go through the people most embarrassed by it, and it only got out because one reporter would not give up. Now think about the channels that share our judgments today: platforms, search bars, and AI assistants that answer in smooth sentences. Many of these are owned by the same companies whose products we are judging. Here, channel means the path a message must travel.
The cable office still stands between every witness and the world. Someone owns that connection, including the AI systems that now shape it. The connection is the medium that can slow, shape, or block what gets out.
Know who it is.
And that is Wisdom Speaking.
Sources
Webb Miller, United Press dispatch datelined Dharasana, 21 May 1930, as archived at UPI Archives.
Wikipedia, “Dharasana Satyagraha” (Gandhi’s arrest on 4 May 1930; the censorship episode; Senator Blaine’s reading of the dispatch; secondary reporting, so check against Miller’s I Found No Peace, Simon & Schuster, 1936). The Leader, “Cass County journalist’s work had global impact”, gives the figure of more than 1,000 newspapers.
John XXIII, Mater et Magistra (15 May 1961), paragraph 236, as presented at Vatican II Journey.
Marshall McLuhan and Eric McLuhan, Laws of Media: The New Science (University of Toronto Press, 1988), on the four questions of the tetrad. Marshall McLuhan, Understanding Media: The Extensions of Man (McGraw-Hill, 1964), text at hannemyr.com. Marshall McLuhan and Quentin Fiore, The Medium Is the Massage (1967), p. 74, via The Marshall McLuhan Dictionary of Quotations.
Multifamily Dive, “RealPage wins pause on New York’s algorithmic rent-setting ban” (2 October 2026).
TMCnet Insight, “Synergy Research Finds AI Lifted Q2 Cloud Revenue to $143.4 Billion” (5 August 2026), and Cloud Tech Daily, “Google Cloud’s share hits record high of 15%, Amazon AWS drops to 28%” (market shares as cited from Synergy; secondary reporting).
Futurum Group, “AI Capex 2026: The $690B Infrastructure Sprint” (12 February 2026; an analyst estimate of company guidance, not a filing). OECD, “International aid fell sharply in 2025, says OECD” (April 2026; preliminary data).
Challenger, Gray & Christmas, “Job Cut Announcement Report, July 2026”.
Wikipedia, “2023 Writers Guild of America strike”, and Brookings, “Hollywood writers went on strike to protect their livelihoods from generative AI” (contract terms).
Mahatma Gandhi Foundation, “When Mahatma Gandhi was welcomed by textile mill workers of Lancashire” (the boycott’s effect on Lancashire mill workers).
Part one’s sources cover Lewis Hine, Addie Card, and the See stage of the method.