AI – Erhardt Graeff

AI’s Humility Problem (fPET 2026)

11 June 202619 June 2026Erhardt GraeffLeave a comment

Citation

2026. Graeff, E. “AI’s Humility Problem: Threats to the Practice of Design.” Presented at the 2026 Forum on Philosophy, Engineering, and Technology (fPET 2026), University of Maryland, College Park, MD, Jun 11.

Presentation Slides

Abstract

This paper will examine the relationship between artificial intelligence and humility, arguing that humility—especially epistemic humility—is a crucial civic virtue that contemporary AI systems place at risk. Generative AI technologies allow us to use knowledge that is beyond us without helping us appreciate the boundaries of our understanding. While AI could help reveal our limitations in ways that augment humility, more common uses threaten to erode humility, the social practices that sustain it, and our civil society that relies on it.

I am approaching this concern through exploratory engagement with multidisciplinary scholarship on humility, ethics, computing, and AI, alongside reflection on my formative experiences in computing and my current work in engineering education. Drawing on my graduate education within the innovation culture of the MIT Media Lab, I will consider, from this situated vantage point, how many computing environments have rewarded anti-humble performances of speed, certainty, and mastery. Generative AI intensifies these tendencies by giving users the illusion that they need not be limited by their own experiences and education—that one can access collective knowledge on demand, even though this is far from the totality of human knowledge. I will use these experiences to ask how everyday encounters with AI may reshape dispositions toward doubt, listening, and deference to others.

A guiding premise is that humility is not merely a private moral trait but a foundation for democratic and collaborative life via openness to plural forms of knowledge and deliberative capacity. Humility ensures that we value the creation of new knowledge, that we are awed when others do things we cannot or did not think to do, and that we embrace curiosity and deep listening. Awareness of our limitations enables us to be more open and tolerant, to collaborate with people from different backgrounds, and to become well-rounded humans. If generative AI obscures our lack of knowledge and ability, I fear we will diminish a key part of our humanity and civic capacity.

I am exploring these questions through a review of literature on intellectual humility, AI and engineering ethics, engineering and computer science education, and critical approaches to human-computer interaction, with strong influence from Shannon Vallor’s account of “technomoral humility” and her arguments in The AI Mirror. Rather than claiming a settled literature, the paper will map concerns about how AI mediates experiences of competence and ignorance. It will also consider the responsibility of developers and educators to account for what happens when humility is undermined and these effects operate at scale.

Ultimately, I contend that AI strengthens the need to cultivate technomoral humility. Particular attention will be given to implications for undergraduate engineering and computer science programs. We need engineers and technologists who see humility as a virtue in their work. The paper will argue for an AI ethics more centrally concerned with humility and for pedagogical directions that normalize admitting uncertainty, foreground the social origins of knowledge, reward engagement with unfamiliar perspectives, and connect humility to civic-mindedness and democracy.

The AI Bargain

2 April 202619 June 2026Erhardt GraeffLeave a comment

Citation

Graeff, E. 2026. “The AI Bargain.” In Anderson, Janna, and Lee Rainie, eds. Chapter 9: Epistemic Vigilance: Discerning Truth, Illusion and Misinformation. Building a Human Resilience Infrastructure for the Age of AI. Imagining the Digital Future Center. https://imaginingthedigitalfuture.org/reports-and-publications/human-resilience-in-the-age-of-ai/.

Link

Epistemic Vigilance: Discerning Truth, Illusion and Misinformation

Essay: The AI Bargain

The AI bargain: AI will be ‘just good enough that we won’t give it up.’ Human resilience requires epistemic humility, cultivating practical reason and investing in humans’ special moral capacities

Artificial intelligence will play a far more significant role in shaping our decisions, work and daily lives over the next decade, not because most people will demand such a transformation, but because AI will be subtly integrated into nearly every digital system we rely on. Even if many of us feel uneasy, resistance will struggle to compete with the promise of efficiency, personalization and productivity. Powerful forces of capital and the lure of perceived convenience may end up deciding for us.

At the moment, there is little appetite for the kind of regulation that might slow this integration. Generative chat assistants are celebrated as helpful companions for writing, coding and learning. Evidence is emerging, contested but concerning, that these tools can undermine attention, learning and even mental health, but the positive press is loud enough to muddy any call for restraint. Protecting children and human resilience more broadly would require moral courage from educators, technologists and policymakers.

We may see pockets of refusal. Elite families already limit screens and social media for their children, while the rest of society is nudged toward greater dependence. But opting out will not be realistic for most people. Technology companies, eager to justify their massive investments in AI infrastructure, are embedding it into learning management systems, workplace software, financial services and everyday tools like email and word processors. Software has long been engineered to be feature-rich rather than fail-safe; AI will amplify that tendency. There will be lawsuits over errors and harms, but large firms will shield themselves behind terms of service and the sheer complexity of their systems. The technology will be just good enough that we won’t give it up.

The AI bargain is no bargain

“This AI bargain comes at a potentially staggering price. In her book ‘The AI Mirror,’ philosopher Shannon Vallor cautions that we are trading something essential when we rely on AI: the ‘space of moral reasons.’

“Democracy depends on our ability to explain and contest decisions, to ask why a loan was denied, a student was flagged or a medical treatment recommended. Yet the deep-learning models powering today’s AI are intrinsically opaque. Vallor, echoing Frank Pasquale’s vision of a ‘black box society,’ reminds us that when reasons disappear behind algorithms, accountability follows.

“The danger to human resilience is not only technical or procedural; it is fundamentally moral. If we cannot meaningfully discuss automated decisions, we will more often than not accept them and grow reliant on them. Vallor warns us about ‘moral deskilling.’ Just as GPS has eroded our ability to navigate with a map, AI may erode our capacity to deliberate, to imagine alternatives and to take responsibility for collective choices.

“If we aren’t cultivating our moral skills in schools, workplaces and civic life, we will erode the practical wisdom that undergirds our human adaptability and resilience. Overreliance on machines risks shrinking our moral imagination precisely when we need it most.

How, then, should we respond?

First, we must cultivate epistemic humility. AI systems speak with unwarranted confidence and humans are tempted to mirror it. Resilience requires the opposite habit: awareness of what we do not know, curiosity about others’ experiences and respect for forms of knowledge that cannot be reduced to data. Schools and workplaces should reward slow reasoning, explanation and disagreement, not just correct answers produced fastest.

Second, we need to maintain social practices that keep the space of moral reasons alive. We should be designing AI systems that show their work. We must create and advocate for more face-to-face human forums in addition to today’s classrooms, juries and community meetings. Automated recommendations should be treated as starting points rather than verdicts. And AI can also be designed and used to reinforce human deliberation. Recent experiments in participatory city visioning in Bowling Green, Kentucky, as well as the large-scale, online deliberations run by Audrey Tang and Taiwan using pol.is, show that AI can widen participation rather than replace it when the design goal is collective reasoning instead of automation.

Third, we should invest in capacities that machines cannot replace: empathy, moral imagination, collective problem-solving and the patience to sit with uncertainty. These are not soft add-ons to technical skill; they are the infrastructure of democratic resilience. If we teach students to use AI and to code AI, we must also teach them when not to automate.

I hope my worries prove overstated. I also fear the kind of cataclysmic failure of an AI-based technology that may shake us out of our complacency. Absent such a unifying event, our adaptability as a species will do what it always does.

Technology, when embraced, always transforms human decision-making, work and daily life in some way. We risk degrading the moral skills and practical wisdom required for decision-making, creativity, self-care and social life until these capacities begin to feel impossible without AI assistance. The AI bargain is not settled. Let us defend the fragile, human space where reasons matter and design technologies that serve that space rather than replace it.

The AI Mirror book review

2 July 20253 July 2025Erhardt GraeffLeave a comment

The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking by Shannon Vallor
My rating: 5 of 5 stars

My fellow technologists, policymakers, educators, and education leaders wrestling with the impacts of generative AI should read Shannon Vallor’s excellent book The AI Mirror as soon as possible. In this highly readable and useful work of philosophy, the virtue ethicist Vallor calls for reclaiming our humanity in an age of machine thinking through moral wisdom and prudence.

The book starts with two organizing concepts. First, the metaphor of AI as mirror is carefully constructed to help explain how the current generation of AI technologies operates. They reflect back what is fed into them. They have no sense of the world; they inhabit no moral space. Of course, humans can’t help but anthropomorphize technologies that have human-like behaviors—projecting onto them reasoning abilities and intentions. There is a long history of this, and it’s used as a design pattern in technology to enhance usability and trustworthiness. But this is a trap. Machine thinking should not be mistaken for a machine caring about you or making moral decisions that weigh the true complexity of the world or a given, specific situation. Generative AI predicts the kinds of responses that fit the pattern of content it has been trained on.

Vallor’s other conceptual starting point comes by way of existentialist philosopher José Ortega y Gasset, who suggested that “the most basic impulse of the human personality is to be an engineer, carrying out a lifelong task of autofabrication: literally, the task of making ourselves” (p. 12). Vallor worries about how our future will be shaped if we rely on a tiny subset of humanity to design and build our AI tools—tools based on a sliver of the human experience—and we then rely on those reflections of biased data, filtered by the values of their creators, to guide society via AI-based problem-solving and decision-making.

Explaining why this is such a big problem is helped by Vallor’s use of another metaphor, “being in the space of reasons”, which describes “being in a mental position to hear, identify, offer, and evaluate reasons, typically with others” (p. 107). She uses this to contrast AI possessing knowledge with the psychological and social work necessary to make meaning through reasoning. This is not how machines think. “One of the most powerful yet dangerous aspects of complex machine learning models is that they can derive solutions to knowledge tasks in a manner that entirely bypasses this space,” writes Vallor (p. 107).

Furthermore, the “space of moral reasons” represents not only the private reflective space for working through morally challenging dilemmas to arrive at individual actions, but also the public spaces for shared moral dialogue. This is politics. As Vallor notes, “the space of moral reasons is [already] routinely threatened by social forces that make it harder for humans to be ‘at home’ together with moral thinking” (p. 109). AI threatens our moral capacity by seeming to “offer to take the hard work of thinking off our shaky human hands” in ways that appear “deceptively helpful, neutral, and apolitical” (p. 109). We are on this slippery slope toward eroding our capacity for self-government. Technology can trick us into believing we are solving our biases and injustices via machine thinking, when in fact we are reinscribing those biases and injustices with AI mirrors.

Like any mirror, humans will inevitably use AI to tell us who we are, despite their distortions. Social media algorithms do this every day. “For you” pages on TikTok reflect a mix of our choices, our unconscious behavior, and the opaque economies and input manipulation tuning the algorithm. But is this who we are? Is this who we want to be? At our fingertips, with no human deliberation required, we might casually assume the reflection we see is a fair rendering of ourselves and the world. Vallor distills this threat by writing, “when we can no longer know ourselves, we can no longer govern ourselves. In that moment, we will have surrendered our own agency, our collective human capacity for self-determination. Not because we won’t have it—but because we will not see it in the mirror” (p. 139).

One of the reasons I like Shannon Vallor and her writing is that she is not simply a critic of technology. She loves technology. She wants it to work for us. And she spends time in this book describing the ways generative AI can be useful. Large language models perform pattern recognition on data so vast it would take millennia for a human to encounter let alone comprehend, which allows us to learn things about how systems work and find information and connections beyond the reach of mere human expertise. We are already unlocking scientific discoveries with AI that serve humanity.

Vallor encourages us to reclaim “technology as a human-wielded instrument of care, responsibility, and service” (p. 217). Too much of our rhetoric around AI is about transcending or liberating us “from our frail humanity” (p. 219). Replacing ourselves or our roles in self-governance and as moral arbiters will lead to magnifying injustice, making the same mistakes again and again (e.g., racist legal proceedings, sexist health diagnoses) with greater efficiency. We could be using these technologies to interrogate our broken systems and let us fix them, rather than supercharging them. The chief threat of AI is that we will come to rely on it to make morally challenging decisions for us, and the more we do this, the more we erode our individual and collective ability to exercise our moral agency, leaving AI to govern us with a set of backward and inhumane values.

My favorite part of the book is “Chapter 6: AI and the Bootstrapping Problem.” Here, Vallor returns to her arguments in her brilliant 2016 book Technology and the Virtues (my review) and renews her call for the cultivation of technomoral virtue to help us reclaim our humanity amid the din of AI boosterism. In The AI Mirror, she directs her call to my students—the engineers and technologists who will be tasked with building and using AI technologies. I have been writing for years about the need for a renewed professional identity for engineers and technologists that fully embraces their civic responsibilities. This is what drew me to Vallor’s work originally, and it is exciting to hear our calls echo one another.

She takes issue with Silicon Valley’s emphasis on perseverance as a virtue and the technological value of efficiency. If we allow our technology creators and their products to promulgate such values, we risk dooming ourselves to a less caring, less sustainable, less just future. There are some things we should stop doing. There are some applications of AI that we should refuse. And we need virtues of humility, care, courage, and civility to guide us toward moral and political wisdom. We should no longer allow “the dominant image of technical excellence and the dominant image of moral excellence to drift apart”—”neither alone is adequate for our times” (p. 179).

View all my reviews