Humane Studies at 50

IPHS 50th Homecoming Seminars

Katherine Elkins · Saturday, September 26, 2026

We began by returning to Goodhand’s original questions and then put Leopoldo’s work on technology and democracy beside Yuval Harari’s claim that democracy may rely on an outmoded idea of human nature. Democracy assumes that people can form a judgment and act on it. The pressure of the morning was that technologies are getting better at reaching beneath that judgment. They can offer new freedoms, but they can also hack the person who is supposed to be free.

Leopoldo’s first move was to bring us back to Venezuela. For many people, AI and the internet are not the central fact of life; some still do not have running water. I mentioned the current claim that AI could leapfrog ordinary access to the internet, and an alum thought this may indeed happen. But that did not make the earlier point disappear. A technology can reach people without changing the conditions that make a life livable.

Leopoldo also said that a great deal of human experience is missing from the web. Our analyses of social media and web-based NLP can look like analyses of humanity, but much of humanity is not there. This is part of what makes the Schmidt Sciences project so difficult. The challenge is not only to gather more information. It is to find ways of attending to experience that has never been made into data, or that cannot safely be made public.

We then moved to the problem of values. We talked about sovereign AI and whether systems in different countries should reflect very different histories and commitments. Current systems often carry liberal, Western, and American assumptions even when they appear neutral. At the same time, our research has shown that a model may give different answers when the same question is asked in different languages. That did not give us a way out. It made the question of the Good more pressing. Is it the same for everyone? If not, who decides which differences are worth preserving?

I also talked about AI safety claims and the difficulty of auditing them. If a company says its system is safe or aligned with human values, but does not give researchers access to examine it, what exactly can the rest of us know? The question was not only whether a model had values. It was who had the power to choose them and who was allowed to look.

Later, we considered the other side of the technological argument. Smaller models might run directly on a phone. A person might have a chosen assistant, a kind of sage in the pocket, and use it to choose sources rather than simply receive a platform’s feed. This might matter in places without dependable internet access. It was an appealing thought. But it also raised a problem that stayed with us: how many people will really make the work of building a good intellectual world for themselves? A personal assistant can give someone more control and make it easier to live inside a world that answers only to them.

Leopoldo’s story about Twitter made the argument about technology and democracy very real. Twitter had been an opening when he was shut out of television and radio. It let him speak and organize politically. Then the regime learned to use the same space. Bots, lies, threats, and personal attacks made the platform a battlefield. We talked especially about the attacks on women. When public speech becomes dangerous in a particular way, women leave the platform, and the visible public becomes more male. I pointed out that X’s restrictions on access have made it impossible, or far harder, to repeat the kind of research that had followed thousands of tweets. The company controls the public space and now much of the evidence about what is happening inside it.

An alum with therapeutic experience said something that put language to what was at stake. Human beings are much easier to manipulate emotionally than we like to think. We often respond before we know why, and give ourselves the reasons afterward. They described the difference between therapy and manipulation. In therapy, someone may be helped to approach what is painful and return from it, with consent and in the service of greater agency. A platform, political campaign, or AI system can learn the same vulnerabilities and use them against someone. I asked what this means for teaching. Teaching changes people too. It asks them to question their assumptions. The distinction matters.

Raúl brought up David Foster Wallace’s This Is Water. He used it to get at the “I” and the “we.” Wallace’s point was that the default setting of the self can make other people disappear. It gave us a different way to talk about a feed made only for one person, and about the kind of shared world democracy needs. I had been thinking about this in my 2024 Baccalaureate address, “The Shape of a Life”, which returns to Wallace’s speech.

Leopoldo then spoke about preparing for prison and for solitary confinement. He made a daily routine around exercise, prayer, reading, and writing. He said that he learned to pray for gratitude rather than freedom. When hooded guards came into his cell with guns in order to frighten him, he worked on bringing his heartbeat down. That ability to control a bodily response was one small thing that remained his own. I asked whether change sometimes requires periods of isolation, and how someone does not lose an inner freedom in the process. I was thinking of Martin Luther King Jr. in prison, and of my uncle Hank, who brought him books personally.

An alum pushed back on any easy admiration of solitude. In trauma work, the things that help someone survive can harden into armor. They brought up Hector’s helmet in The Iliad: it makes him formidable in war, but frightens his child. Leopoldo did not present isolation as a good life. The force of his story was that a person can find a measure of freedom under terrible conditions. The alum’s point was that survival still has to lead back toward connection.

After the break, Lisa’s Plato seminar felt like a continuation rather than a new subject. She talked about how Plato began interested in science as well as philosophy, but eventually came to think that science was downstream. I loved that phrase as a way of describing what IPHS does. Before we decide what to measure, build, or study, we have already made some decision about what matters.

Lisa used the example of recognizing a dog. We do not begin with a neutral pile of colors and shapes; we see a dog because we have ways of seeing that already contain purposes. An alien with other purposes might see the same animal differently. Jon kept bringing medicine, neuroscience, and AI back into the discussion. Science can tell us an enormous amount about health and the brain, but it cannot tell us why health is good or what human life should be for.

I asked the question that had been following us throughout the morning: if we can question everything we think we know, what grounds our understanding of the Good? I said that some people in the room relied on their spirituality to ground their belief in the Good. Lisa did not make faith, or a metaphysical ground, do that work for her. The important point was that neither science nor AI can settle the question for us.

Over lunch, people spoke about their own faiths and spiritual traditions. Alumni also talked about experiences they had lived through, or witnessed, that they did not think could be explained by pure materialism. We touched on older ways of thinking about reality in which consciousness or information might be more basic than matter. These were not simply theories for the people speaking. They were part of how they understood their lives.