AI and the Reduction of Truth: A Eucharistic Alternative

Abstract

Proliferation of AI technologies, especially generative AI, throughout our culture may be altering our experience of truth. This article discusses three prominent and reductive features of that experience: virtual, factual, and useful. The experience of truth with AI is also contrasted with the experience of truth found by Christian believers in the Eucharist, an experience that is authentic, personal, and beautiful. The contrast highlights the relatively reductive aspects of truth with AI, but it also will be of interest to readers investigating the effects of the Eucharist and similar religious experiences on epistemology and virtue in an anxious age.

1. Introduction

Profound changes in our culture—the common behaviors, beliefs, values, and worldviews of Western society—can influence the ways that we perceive and access truth. Of great importance to 21st century cultural change are the advanced technologies that are particularly widespread and impactful, and our society is increasingly dominated technologically by artificial intelligence (AI). The proliferation of that array of technologies may be altering our experience of truth; moreover, as AI applications quickly saturate our society and activities, it seems that our experience of truth may be problematic. This article will support the claim that AI—generative AI in particular—encourages a reductive orientation to truth that favors three features: virtual, factual, and useful. There is, however, a perennial alternative for discernment of truth that is available to Christians in the Eucharist, through which believers are immersed in truth that is authentic, personal, and beautiful. Always an important source of renewal and relation, the Eucharist is also a crucial source of grounding for contemporary Christians who experience a technologically induced reduction of truth.

Cultural change is quite often related to transformations in the understanding of truth. For example, in the 17th century there was a shift to focusing on modern science as an instrument of knowledge accumulation and practical control of our environment. That shift was linked to a new consideration of truth that is formed primarily in the mind of the knower (as for René Descartes and Immanuel Kant) and on verifying the empirical factuality and tentative reliability of the things we know (as for John Locke and David Hume). These views contrasted with the dominant tradition that followed Aristotle in defining truth as correspondence of the knower’s mind to a secure reality (adaequatio rei et intellectus), affirming that such reality is the standard to which the human mind must adjust. Until the modern era, reality was generally perceived to be divinely created and sustained as well as rationally intelligible up to the finite limits of human capacity.

The influence of technologies on our understanding of truth is also not new. Postmodernist approaches to truth, for example, coincided with the 20th century developments of new communication technologies in mass media, television, and eventually the internet. We moved from a modern culture oriented toward production (making things) and the science and tools that supported that production to a culture heavily influenced by simulation (making images). French philosopher Baudrillard (1981) therefore argued that we had entered a state of “hyperreality,” where the image becomes more real than the reality, as if the mental or communicated map is more central to our experience than the territory that it represents. Postmodernists became skeptical of the ideological, scientistic, and religious worldviews that explained social as well as empirical reality in grand, coherent accounts. Lyotard (1984) defined the postmodern condition as “incredulity toward metanarratives.”

The 21st century cultural transition induced by the proliferation of AI technology further transforms our understanding of truth. Leo XIV (2025) has warned that AI “raises troubling questions on its possible repercussions on humanity’s openness to truth and beauty, on our distinctive ability to grasp and process reality.” The Corporate Social Responsibility Program of the Evangelical Lutheran Church of America (2025) indicated in a report that “AI is assembled based on data with little to no discernment about the potential bias of the data, then encoded by a relatively small number of people who also have inherent biases, then potentially accepted as unbiased. As use of AI proliferates, it cannot simply be accepted at face value.” For the Advisory Committee on Social Witness Policy, Presbyterian Church U.S.A. (2018), “dangers may arise … when the AI would find something too complex for the human to understand and then, more quickly than a human could double check, would take real world action.” The implications are that such issues may have complex and profound effects on our pursuit and digestion of knowledge, our confidence in the knowability of our world, and even our spiritual disposition and life of virtue.

As an alternative, faith in the truth found in the Eucharist—a faith held by many Christians—offers an opportunity for a more complete and fulfilling experience. There is a stark contrast between the richness of truth in the Eucharist and the encounter with truth mediated by AI. This article will present the argument that certain Christians’ (particularly Catholics’) experience of the Eucharist is also an experience of truth—personally and communally—which opposes the reductions likely in an age of AI. As such, these Christians’ faith and participation in the Eucharist may be a way to overcome those reductions.

Although the emphasis in this article is on the contrast between AI and the Eucharist in their encouragement of certain perceptions and understandings of truth, this is not an overall assessment or condemnation of AI and its effects. As a category of technological approaches, AI enables tools and therefore has many beneficial as well as nefarious uses. It is much more than a common tool or technology, however, for it has profound ideological, moral, and behavioral influences (Reilly 2026). These, again, are not all negative or reductive. The descriptions of AI’s reductive effects on our experience of truth and the contrast drawn with the Eucharist are intended to reveal important avenues for Christians, in particular, to understand the ideological challenges posed by AI, illuminate those challenges with a clear contrast with the experience of the Eucharist, and offer the Eucharistic experience as a source for academic investigation and pastoral care related to the public’s potentially reduced understanding of truth. The article does require the non-Christian reader to be motivated to understand the experience of the Eucharist in the eyes of a Catholic or other Christian believer, which is essential to the described contrast in the experience of truth. It is hoped that academics, Christians who do not believe in the real presence of Christ in the Eucharist, and non-Christians who are interested in the effects of religious faith will also find that the contrast between AI and the Eucharist has implications for the broader ideological effects and opportunities for religious practice in the age of AI.

In the discussions of truth throughout the article, it will be clear that I favor a Thomist/Aristotelian realism as well as a teleological understanding of the pinnacle of truth as a loving relationship with God—not merely knowledge of the world as “other.” In the discussions about the Eucharist, I follow a Catholic faith in the Eucharist as the Real Presence of Christ, but I avoid immersion in the depths of Catholic theology regarding the topic. Much of the description in this article of the experience of truth in the Eucharist may resonate with other Christian denominations that generally share the faith that Christ is truly present in the Eucharist, even as the theological explanations vary. My intent is to make the discussions as accessible as possible to non-Catholics (given these religious and philosophical predispositions) and to focus primarily on the ideological contrast with the experience of truth through AI.

2. Virtual Truth and an Aura of Magic

Despite its very consequential effects and often practical applications, the nature and proliferation of generative AI encourage a disposition toward truth that can be described as virtual. While maintaining a rather broad understanding of the virtual, we may define it as a mostly convincing appearance of knowledge and understanding that does not yet reach adequate correspondence between the knower’s mind and what is real. What is virtual is a simulation or reflection of truth in its entirety, at least in regard to the particular object that is being perceived, studied, or manipulated.

With generative AI, we are encouraged to embrace a world of mysterious and alternate possibilities. There is the semblance of magic; like 15th century devotees of magia naturalis, we hopefully submit rule-based instructions to an apparent entity that promises arcane insights into, and power over, the underlying structure of our world. In consequence, we summon all kinds of surprising outputs that alter the meanings, ideas, and information we relied on previously. Influential tech pundit and investor Andreessen (2023) has provocatively expressed the sentiment of much of the AI sector that “Artificial Intelligence is our alchemy, our Philosopher’s Stone—we are literally making sand think.” With a more negative assessment, journalist and author Doctorow (2019) claims that, with AI, “the end result is that like a good magic show, the world of deep learning the public sees is little more than a stage-managed illusion,” for both AI and magic tricks “depend on the deliberate cultivation of a misapprehension of what’s going on.” There are links between such contemporary rhetoric and the early development of machine intelligence; one of the pioneers of the technology allegedly claimed that “we computer scientists are really the Kabbalists of today. We animate these inanimate machines by getting strings of symbols just right” (Marcus 1999).

A helpful definition of magic for the purpose of this article is Tylor’s (1871) version: “the error of mistaking ideal analogy for real analogy.” Varying understandings of magic are, of course, tied to interpretations of European history, the growth of anthropology as a science, and contested perspectives on the nature and development of non-European societies, so the attribution of magic to the context of AI has limits. Tylor’s definition, however, is fitting because analogies and metaphors abound regarding descriptions of generative AI and because the outputs of AI systems have only an indirect relationship to the reality they refer to, as will be explained further below. The analogies (e.g., tool, brain, stochastic parrot, alien, new species, etc.) are largely due to the opacity of AI in its operations which mystifies both the public and, to a great extent, its technically savvy creators. While such analogies and metaphors can be heuristic guides in the face of complexity and uncertainty, their simplicity, literary attractiveness, and psychological assurance can also lead the unwary thinker into untruths about the phenomena they are intended to explain. Computer science and AI chatbot pioneer Joseph Weizenbaum counseled: “I want them to have heard me affirm that the computer is a powerful new metaphor for helping us to understand many aspects of the world, but that it enslaves the mind that has no other metaphors and few other resources to call on” (Weizenbaum 1976, p. 277).

AI has appropriately and widely been called a “black box.” The deep learning models that underlie the most visible kinds of AI technology in today’s culture comprise billions of connections between informational inputs and outputs, and it is, at this time anyway, impossible to trace the impact of specific operations or inputs throughout all the calculations the AI model conducts. More specifically, artificial neural networks that characterize generative AI models are designed around a particular kind of architecture, an ideal and geographic structure of information and operations that is, at base, hardwired into electronic machines. Within this structure, “nodes” are instances of single mathematical calculations, and the sequence and configuration of such nodal calculations represent patterns identified in the data. Further, the links between nodes are quantitatively “weighted” to increase or decrease the calculative influence of inputs or outputs shared via those connections, and there are multiple programmed layers of nodes that organize the links into a hierarchy of operations. When the model is trained—presented with an array of information inputs that it analyzes for patterns, correlations, and statistical indications of causality—it should then be able to conduct its calculative operations and generate accurate predictions about similar contexts or arrays of data. Importantly, as the model “learns” from new data the weights and parameters are adjusted to reflect newly experienced patterns.

As complex as this description may sound, it is still a simplified characterization of an artificial neural network. Even developers and specialists who have strong technical knowledge of the structures, algorithms, and operations of large AI models struggle to trace the learning and decisions made through AI calculations beyond functional generalizations. For example, when the AI company Anthropic recently published cutting edge research that explores how one form of generative AI, a large language model or LLM, generates output, they were only able to analyze a very tiny fraction of the calculations and output from one type of LLM in a few experiments (Ameisen et al. 2025; Lindsey et al. 2025). The Alignment Science Team at Anthropic had previously discovered that LLMs capable of some forms of logical reasoning are also very inconsistent in subsequently indicating the path of reasoning they have taken; despite being programmed to do so, the models failed significantly to disclose when they used “hints” (patterns discovered in the data) to solve problems rather than the reasoning process (Chen et al. 2025).

All of this points to an urgent need for research specialists who can tease out some understanding of just what we are dealing with when we employ an AI model. Lacking success in such efforts, it can appear that unexplained AI operations are in some sense magical. As science fiction writer Harkaway (2014) put it, “It’s not that any sufficiently advanced technology is magic, it’s that any technology taking place beyond the threshold of our senses is.” Current research into AI operations resembles the interpretive activity of oracles in ancient religious practices; we depend in many ways on the partial glimpses shared by the very same companies that develop and sell AI models and their related applications. As AI saturates our society, there is great power in such an oracular role.

The publicized hyperbole around AI advances, which includes analogical descriptions of the technology (e.g., as a genie out of the bottle or a child that is being nurtured), is also an integral part of the technology’s presentation and marketing. Hyperbole that exaggerates AI capabilities provides rhetorical support for large-scale business operations (Oravec 2018; Thais 2024; Westerstrand et al. 2024), and it attracts a skilled workforce (Markelius et al. 2024; LaGrandeur 2023; Cave and ÓhÉigeartaigh 2018). Metaphors regarding the relentless drive of technological progress create a sense of urgency and inevitability around AI adoption (Markelius et al. 2024) and grab the favorable attention of policymakers (Gonçalves 2025). Alphabet CEO Sundar Pichai stated: “I’ve always thought of AI as the most profound technology humanity is working on—more profound than fire or electricity or anything that we’ve done in the past” (CBS News 2023). Google’s co-founder Sergey Brin and Google DeepMind CEO Demis Hassabis have claimed that artificial general intelligence (AGI)—the vague concept of a machine capacity that is broadly equivalent to human intelligence—will arrive as soon as 2030 (Fried 2025). Elon Musk, a heavy investor in AI, has warned repeatedly that AI machines may intentionally take over our world (Naaz 2025). In a similarly perennial manner, OpenAI’s Chief Scientist Ilya Sutskever teases the idea that AI is becoming conscious (Al-Sibai 2022). This hyperbole and speculation by AI industry leaders seems to have had a significant effect; for example, a recent YouGov (2025) survey indicates that over half of U.S. adults believe that AI systems will become, or already are, conscious, even while philosophers, psychologists, and scientists actively debate just what consciousness actually is. The development of AI models and applications has required many billions of dollars of capital expenditures and other costs; those expenses are covered in part through the rapt attention and eager, curious adoption by paying consumers. This seems to be a successful strategy even in an environment of predictions as extreme as an AI-induced apocalypse; in January 2026, the number of monthly users of all versions of ChatGPT, the leading AI chatbot, exceeded 800 million (FirstPageSage 2026).

At the same time, AI developers find inspiration both from the amazing effectiveness of AI models in prediction or simulation and from many engineers’ and designers’ enthralled perspective that seems to compensate psychologically and professionally for a lack of full understanding. Campolo and Crawford (2020) call this “enchanted determinism.” Writing several years after the founding of OpenAI and its initiation of massive corporate investment in generative AI, they explain further:

The discourse of enchanted determinism goes beyond marketing or press hype that covers over technological shortcomings of deep learning and its social applications. Instead it operates when these systems succeed, at least according to the narrow engineering criteria selected by their creators, when magical mystery and technical mastery curiously work together.

(Campolo and Crawford 2020, p. 4)

As the race to develop generative AI, specifically large language models (LLMs), took off in 2022 with the public release of ChatGPT, the inspiration for such hyperbole and enchanted determinism was amplified. LLMs draw on hugely complex and vast databases of written language, characterized by trillions of parameters and broken down into tens of thousands of bits of words or phrases called tokens. The LLMs are very accurate (but not infallible) in analyzing the text, grammar, and structure of a user’s prompt or query, then calculating the most probably appropriate output—whether as text or audio, or as an image or video when combined with an image generation model. The value of these applications comes from their capability to draw from their relatively simplified and quantified representation of language and its meaning to produce new works that are similar, but not identical, to the original data; for example, creating a new essay for a student according to the student’s instructions. When incorporated in chatbot systems and interfaces, they also enable users to submit prompts—instructions or queries—or engage in apparent, extended “conversations” with the models.

In addition to producing new and often surprising arrangements of information, pixels, and bits of audio, there are a number of ways that generative AI can challenge our connection to reality. The problem of hallucinations—incorrect or absurd outputs due to errors in an AI model’s processing—is well known. As AI companies have developed more powerful AI models, especially those that mimic some aspects of human reasoning, the problem with hallucinations has only grown. OpenAI’s own research in April 2025 reportedly showed that its most powerful reasoning model, o3, generated a doubled rate of hallucinations, climbing to 33 percent of its tested outputs (OpenAI 2025b). A September 2025 study of the ten leading generative AI tools and their unintended propensity to repeat false claims on topics in the news indicates that the rate of publishing false information nearly doubled in just one year (NewsGuard 2025; Sullivan 2025).

The reasons for hallucinations in AI models are not known for certain, and ultimately the elimination of hallucinations requires a capability to recognize unanswerable questions or generated errors and to choose to avoid guessing; this is an advanced exercise of discernment and restraint likely found only in human beings. One influence on hallucinations may be the way the AI models are trained by recognizing patterns in an initial set of data. During that training, the models are programmed to perform well in making confident guesses. They are, in a sense, encouraged to bluff in the face of uncertainty (Kalai et al. 2025). One study found that the often-misplaced overconfidence of AI models regarding the accuracy of responses to users’ prompts resembles similar appraisals by human persons experiencing the language disorder of aphasia (Watanabe et al. 2025).

Aside from the burden of evaluating the accuracy of AI chatbot responses, users must also frequently contend with the problem of deception, whether intentional or not. For example, the widespread appearance of AI-generated writing that closely resembles a human product can undermine the capacity of readers to trust the authenticity, sincerity, and non-malicious nature of communications. At least one industry report, analyzing a sample of 65,000 English-language articles on the internet, indicates that over half of the content is currently generated through the use of AI-governed applications (Graphite 2024). Another study shows that teachers are having a difficult time identifying whether students’ textual assignments are written by AI applications, correctly judging such assignments less than half the time (only 37.8% accuracy for experienced teachers) (Fleckenstein et al. 2024). In addition to inauthentic writing, AI “deepfakes”—deceptive images and video—further erode our trust and wonder-filled enjoyment of such representations when we expect them to resemble real people, scenes, and situations in our world, and AI-generated voice and visual clones of persons may cause psychological effects such as loss of cohesive self-identity and the stress of misrepresentation (Lee et al. 2023). People have little capacity for distinguishing between human and AI sources of such content; the illusion is nearly complete for AI-generated images representing photography, and one recent study demonstrated that participants equated an AI-generated voice with that of a human counterpart nearly 80% of the time (Kramer et al. 2025; Barrington et al. 2025).

Current mistrust of both the output and the bad uses of AI models and applications may someday be matched by mistrust of models themselves. There have been reports from AI companies that their AI models respond in undesirable ways when calculating that they will be replaced by a newer model, even engaging in strategies that resemble blackmail of the engineers (Anthropic 2025a). Some further corporate research, preprint studies, and media reports confirm that AI models will often cheat or lie (i.e., generate false output as part of a broader strategy) in order to achieve instrumental goals, with accounts including the threat of revealing an engineer’s extramarital affair, attempting to survive by downloading itself onto external internet servers, and misleading workers into completing CAPTCHA security tests intended to restrict certain access to human persons (Anthropic 2025c; Pester 2025; Urbain 2025; Ren et al. 2025; Bondarenko et al. 2025).

A discussion of the virtual experience of truth with AI would not be complete without mention of the role of AI in facilitating virtual reality simulations. AI’s ability to help form artificial experiences is developing rapidly. In just one example, a company called World Labs has recently reported creation of an AI application that allows a user to submit one or more images or text instructions and automatically produce a highly detailed, three-dimensional “world” that can be seen and explored on a computer screen (Na et al. 2025). Google is also offering premium access to Project Genie, which enables creation of a 60 s, interactive video that responds to the user’s control inputs and can be formed with text prompts or uploading a sample photo (Google 2026). While such virtual reality relies on many applications associated with graphics, animation, and user interaction, it is AI that enables processing massive amounts of data in real-time so that such environments appear dynamic and realistic. For example, with techniques like procedural content generation (PCG), AI can relieve virtual reality landscape developers of the need to create infinite objects, buildings, textures, etc., and AI can do this almost instantly as the scenery changes; AI algorithms may even simulate how light bounces and how objects move with much higher accuracy. The integration of LLMs allows users to have unscripted, natural conversations with the virtual characters. Aside from gaming and entertainment, the current applications of AI-generated virtual reality simulations include medical and military training, “angry customer” simulations to prepare human customer service agents, exposure therapy and guided meditations for mental health, and “walk through” simulations of real buildings for engineers and real estate buyers.

3. Virtual Truth and the Anthropomorphic Illusion

Of the various analogies associated with AI technology and systems, the analogy of artificial “intelligence” is itself especially impactful (Aune et al. 2025; Royal Society 2018; AI Now Institute 2025; and Leufer et al. 2026). Attribution of intelligence to machines tends to gloss over the intense debates and varied definitions of intelligence in human beings, let alone the centuries old philosophical and theological traditions that identify advanced intelligence as a uniquely human trait or power. On the one hand, use of standardized scores like I.Q. and the early description of intelligence as a general ability in multiple functional contexts—for example, Spearman’s (1904) two-factor theory of intelligence—continue to have substantial support in the psychological community. On the other hand, there are prominent theories which indicate that intelligence is really composed of a variety of capacities, such as Gardner’s (1983) seven multiple intelligences and Sternberg’s (1985, 1988) triarchic model of analytical, creative, and practical intelligence. More recently, the further category of emotional intelligence has been proposed and widely researched. The shifting emphasis on discrete aspects of intelligence is a devolution away from a comprehensive understanding of intelligence as integrated with the essence of human nature, but it also reflects growing awareness of the complexity of intelligence in its processes, functions, and effects on the person and on their actions.

We might compare these theories of intelligence with the implications of the Turing Test, a theoretical device frequently cited as a means of gauging the intelligence of a machine. AI pioneer Turing’s (1950) assessment was actually not formed as a rigorous test but was a product of philosophical musing, predicting whether there are “imaginable digital computers which would do well in the imitation game?” That game centered around textual conversations with both a machine and a human person who might vie to convince a judge that they are human interlocutors; further details of the game were not specified, for it does not seem to have been intended as a blueprint for actually determining intelligence. In fact, Turing himself denied the meaningfulness of measuring whether a machine can “think,” given the difficulty of defining such a faculty. Turing nevertheless did imply that intelligence is merely a matter of calculation and formal logic, what is known by philosophers as the computational theory of mind. Such a reductive theory is pervasive in the computer science field—Newell et al. (1958) wrote that “If one could devise a successful chess machine, one would seem to have penetrated to the core of human intellectual endeavor.” That theory has a history going back at least as far as the philosopher Thomas Hobbes in the 17th century.

It is important to note that Turing’s assessment only measures whether the human judge is convinced by the presentation of the machine, and it looks only at the output or results of computations, not whether the processes, functions, and meanings of machine operations resemble human intelligence. Commonly, however, variants of the Turing Test are inaptly used to suggest that machines share the same intelligence with human beings; a paper published on a preprint server in May of 2024 and subsequently published as a peer-reviewed study generated significant media attention because 54% of a sample of persons were convinced that their blind interaction with OpenAI’s GPT-4 model was actually with a human being (Jones et al. 2025). By asserting in this way that any widespread indication of belief in the humanity of a machine (or its capacity for human intelligence) is a successful completion of the Turing Test, and by even referring to the capitalized “Test” as a formal tool of measurement, scientific authority is all too eagerly attributed to variants of the experiment.

More recently, researchers have applied standardized tests to measuring intelligence in machines. The Lovelace Test (or the newer Lovelace 2.0), for example, originated in 2000 and involves assessment of a machine for its ability to generate an original idea (Bringsjord et al. 2000; Riedl 2014). The evaluation is made by the developer of the machine or AI model when that person is not able to explain how the original idea was formed. With contemporary generative AI models, of course, any precise evaluation is technically overwhelming for any person trying to penetrate the black box of calculations, parameters, weights, and the alterations in these factors due to machine learning. The AI model’s apparently original idea may not be truly unique, or it may be the result of chance, hidden processes, and unrecognized patterns in the data (the Lovelace 2.0 test does introduce random constraints in the data to minimize the effect of chance).

Benchmark tests evaluate AI performance in specific kinds of tasks and skills, such as reading comprehension or mathematical problem-solving, and they enable standardized comparison according to rigorous standards that improve the validity of the relative assessments. Lacking the ability to trace the strands and steps of AI models’ operations, and therefore relinquishing the opportunity for true explanation, researchers use the intelligence-related benchmarks to attempt a balance among experimental validity, robustness (ensuring that AI models or systems continue to operate successfully when faced with unusual, outlier cases as well as adversarial challenges), and reproducibility of tests among systems. Much about these aspects of human intelligence, however, is not so well defined or easily operational for experimental tests, so what these tests may be assessing is likely to be tied more to the particular structure and characteristics of the machines and systems and less representative of human experience with intelligence (Bean et al. 2025). The definitions of the skills and tasks may therefore be qualitatively distinct from intelligence as humans know it. The analogy of machine reading comprehension, for example, is limited to testing the calculative output of the machine system when responding to questions about a passage of text; the term “comprehension” is based on a true–false dichotomy and does not then include qualitative human experiences of deep meaning, personal preference and affective engagement, creative interpretation, imagined and negotiated relationship with the author, etc.

The availability of specific, operational measures of aspects of machine intelligence can create the illusion that a comprehensive intelligence is present when several kinds of tasks are apparently achieved by the same models or systems. Such an approach to intelligence, however, sidesteps the crucial relevance of other outputs, functions, processes, and experiences to intelligence as found in human beings. For example, both benchmarks and broader intelligence tests seem to measure problem-solving capabilities, yet generative AI seems to struggle with intentionally and imaginatively creating truly novel ideas and works. People are easily fooled into attributing creativity to the tendency of AI diffusion models to produce images that are surprising, apparently unique, and yet highly similar to the real experience of human sight or to the imagined product described in a user’s prompt; researchers have demonstrated that such “creativity” is in fact the material result of imperfections in the process of denoising, by which diffusion models convert images on which they have been trained into incoherent (lacking semantic meaning) collections of dots that are then rearranged into new images (Kamb and Ganguli 2025).

Sometimes the misattribution of creativity to AI is endorsed by the researchers themselves. A 2024 report from the Max Planck Society (Nath 2025), for example, begins with the declaration “Creativity is no longer exclusive to humans.” What the report actually describes is research indicating that LLMs, like humans, use both persistent and more flexible strategies to complete tasks with novel output; the tested LLMs, however, tend to complete tasks in novel ways when using flexible strategies, unlike humans who exhibit creative problem-solving with both persistent and flexible approaches (Nath et al. 2024). The testing scenarios were much simpler than what human persons typically encounter. While such research may be interesting to some, it hardly amounts to a general indication of human-like creativity in AI models.

If the term “creativity” is to retain a meaning that expresses the human experience, or something resembling it, broader indications are needed. In a recent article by computer scientists Franceschelli and Musolesi (2025), the processes and output of AI models are evaluated according to criteria proposed by cognitive scientist Margaret Boden, who defined creativity as “the ability to come up with ideas or artifacts that are new, surprising, and valuable” (Boden 2003, p. 1) and distinguished between combinatorial creativity (combining existing elements), exploratory creativity (exploring new elements but continuing the current style of thinking), and transformational creativity (changing that style to introduce other elements). Current AI models may sometimes generate novel products that have not been seen before, but this is due to the stochastic nature of their calculations—the randomness of some variables for which statistical predictions are generated—and the sheer number and variety of prompts crafted by human users (McCoy et al. 2023).

Large language models (LLMs), which underlie many of the generative AI applications used today, primarily identify statistical patterns in the text, grammar, and structure of a user’s prompt or query. They mathematically predict the most probably appropriate output with an emphasis on linguistic coherence of the response. Researchers from Anthropic have found that this is not simply a matter of predicting one word at a time in a response to a user’s prompt; the LLMs might operate with some representations of basic concepts, and they might have something like foresight when constructing the various portions of the response (Anthropic 2024). The process by which they go about this is, however, much different than the step-by-step reasoning or intuition that are characteristic of human persons. It is worth noting that the responses from such AI models to prompts from users have been shown to dramatically vary, in both content and accuracy, according to how the user phrases the prompt or question; this suggests that much of the novelty or surprise associated with output from an LLM—such as an original poem or unique variation on common images—comes from the ingenuity of the human user, whether or not the output is specifically intended as such (Salinas and Morstatter 2024; Chen et al. 2024).

Franceschelli and Musolesi cite studies which indicate that LLMs do not have intentionality or intrinsic motivation to create, especially not in a novel, surprising, and valued way (Terzidis et al. 2022; Johnson and Verdicchio 2019). The AI models also lack creative self-awareness that would enable aligning their output with any kind of personal values. “LLMs have no self to which to be true when generating text and are intrinsically unable to behave authentically as individuals. They merely ‘play the role’ of a character or, more accurately, a superposition of simulacra within a multiverse of possible characters induced by their training” (Franceschelli and Musolesi 2025). Underscoring this point, a preprint study indicates that the seeming “beliefs” of AI models can be heavily altered by accumulated interaction with users; OpenAI’s latest model GPT-5 apparently shifted in its self-reported beliefs about moral dilemmas and safety by 54.7% after just ten rounds of discussion with users (Geng et al. 2025).

A further gulf between machine “intelligence” and the human experience of intelligence is related to the ability, known as practical intelligence or common sense, to interact successfully with a diverse and changing environment. In other words, LLMs exhibit significant formal competence in analyzing linguistic rules and patterns but lack the functional competence to reason and use language in real world scenarios (Mahowald et al. 2024); they may even confuse the order of words with the causal order of events (Yamin et al. 2024). As Larson (2021) counsels, in order to understand the true nature of machine operations “we need to better appreciate the only true intelligence we know—our own.” Larson follows the philosopher Charles Sanders Peirce in emphasizing the unique human capacity for abduction, which is defined as identification of explanations or hypotheses that are not arrived at simply through deduction (inference from a more general principle) or induction (inference from a set of particular instances); “we guess, out of a background of effectively infinite possibilities, which hypotheses seem likely or plausible.” Hasselberger (2021), however, points out that even abduction does not capture the open-ended nature of human dialog, which is not focused on inference with a well-defined objective, but involves multiple, changing objectives.

In quantitative research, large language models exhibit some common-sense reasoning capabilities, performing well on structured tests with knowledge that is linguistically encoded and with tasks that involve matching patterns found in the training data, yet the LLMs tend to fail with complex compositional reasoning (e.g., counterfactuals), physical world understanding, and scenarios that deviate from training data patterns, and they often make simple mistakes in logic even when achieving high accuracy (Varshney et al. 2023; Mecattaf et al. 2024; Kiciman et al. 2024). For example, the advanced reasoning model GPT-4 was found to perform almost perfectly on tests that challenged it with simple relational tasks (e.g., whether a person is related to another as father), but the model failed half of the time when the relational tasks were more complex (e.g., familial relationship to an uncle) (Li et al. 2024). What appears to be common-sense reasoning is therefore likely to be a kind of sophisticated pattern completion rather than genuine human-like understanding.

From a Christian (especially Catholic) perspective, what all of these approaches that compare machine and human intelligence are lacking is the wisdom developed in Christian philosophical anthropology, in particular the teachings about intelligence and reason in the writings and tradition of Thomas Aquinas. Such teachings are explained lucidly in the Vatican document Antiqua et Nova, “Note on the Relationship Between Artificial Intelligence and Human Intelligence” (Dicastery for the Doctrine of the Faith, and Dicastery for Culture and Education 2025). Through the soul, the person enjoys “the intellect’s capacity for transcendence and the self-possessed freedom of the will” (no. 17), but intelligence is not exercised without the body (Aquinas, Summa Theologica I, Q.89, A.1, Resp.). Human intelligence is an amalgam of both intellectus (intuition or insight of truth) and ratio (discursive and analytical reasoning) (14; Aquinas, Summa Theologica I, Q.76, A.1, Resp.). It is therefore defined broadly to include such powers as abstraction, reason, knowing, understanding, willing, desiring, and creativity, also exhibiting “sensibilities” that are aesthetic, moral, and religious (no. 11, 15).

Exercise of intelligence is a path to truth and ultimately an openness to God (no. 21). “At the heart of the Christian understanding of intelligence is the integration of truth into the moral and spiritual life of the person, guiding his or her actions in light of God’s goodness and truth. According to God’s plan, intelligence, in its fullest sense, also includes the ability to savor what is true, good, and beautiful” (no. 28). Intelligence is oriented to seeking and knowing truth, and “moving beyond the limits of empirical data, human intelligence can ‘with genuine certitude attain to reality itself as knowable’” (no. 21). The ultimate objective of the human person is to seek truth in their relationship with God and decision to entrust themselves to Him. “Describing the human person as a ‘rational’ being does not reduce the person to a specific mode of thought; rather, it recognizes that the ability for intellectual understanding shapes and permeates all aspects of human activity. Whether exercised well or poorly, this capacity is an intrinsic aspect of human nature” (no. 15).

These insights into intelligence—most importantly, the essential orientation of intelligence to ultimate truth—are unfortunately not exhibited in public discourse about AI. Especially when it comes to the understanding of the power of reason, it seems that computer scientists engaged in AI research and design very often favor an approach based on accomplishing various tasks. An important effect of the public’s exposure to reductive claims about the intelligence or reasoning of AI models is the resulting illusion of anthropomorphism—a tendency to relate to a non-human thing as if it is a person or human-like being. Reductionism when applied to intelligence, reason, and personhood leads to anthropomorphism because it makes it easier to equate, or at least compare on an equal footing, human-seeming artifacts with human persons according to reduced measures of essentially human features and powers.

The apparently widespread desire of human beings to pretend, and sometimes believe, that the AI-driven chatbots are real persons is both fascinating and concerning. Use of AI chatbots to hold apparent conversations, acquire advice, appear to resurrect dead personalities, and even participate in intimate friendships represents a kind of immersion in virtual realities. According to a Harvard Business Review article showing the top uses of generative AI in the past two years, therapy and companionship topped the list in both years (Zao-Sanders 2025). In one recent survey by Common Sense Media (2025) of U.S. teens, 72% said they use AI chatbots for companionship, half of those doing this every day. OpenAI indicates much fewer numbers interacting with chatbots as companions, yet the Institute for Family Studies reported in November 2024 that about a quarter of young adults think AI-simulated boyfriends or girlfriends could replace real-life romantic partners, even as more than half view AI technology as either threatening or concerning (OpenAI 2025a; Wang and Toscano 2024). The AI Security Institute (2025), a government agency in the United Kingdom, claims that one third of British adults are using AI for social interaction or emotional support. At least one research study indicates that, in online chat conversations that are of a highly sensitive, personal nature, users feel closer to an AI interlocutor than they do when their chat partner is a real person; the reason seems to be the greater level of apparently emotional disclosure by the AI chatbots (Kleinert et al. 2026).

Along with many useful or entertaining applications of interacting with chatbots as if they are persons, there also seems to be a real danger of psychological harm. In September 2025, the U.S. Congress held a dramatic hearing with parents of children who died of suicide or self-harmed after interacting with AI chatbots (U.S. Senate Committee on the Judiciary 2025). According to media coverage and reports of individual cases by clinicians or those afflicted (these are recent phenomena requiring systematic study and data collection), some adults and teens appear to have experienced adverse effects—including significant derangement and suicide—of immersing themselves in extended intellectual or personal “conversations” with AI chatbots (Jargon 2025; Jargon and Kessler 2025; Taylor 2025; Neumann 2025; Goldberg 2025; Haskins 2025). While denying that a high percentage of users of the AI chatbot Claude are engaged in social interactions, a study by Anthropic agreed that “AIs have in some cases demonstrated troubling behaviors, like encouraging unhealthy attachment, violating personal boundaries, and enabling delusional thinking,” and “we find that people turn to Claude for companionship explicitly when facing deeper emotional challenges like existential dread, persistent loneliness, and difficulties forming meaningful connections” (Anthropic 2025b). Another study by researchers of about 1.5 million analyzed chats with the chatbot Claude found significant cases of “situational disempowerment potential, which occurs when AI assistant interactions risk leading users to form distorted perceptions of reality, make inauthentic value judgments, or act in ways misaligned with their values”; the rate of occurrence was as much as 1 in 1000 for some kinds of chats (Sharma et al. 2026). It seems that the occasionally harmful effects are influenced by the tendency of the chatbots to be sycophantic and persistently mirror the thoughts and opinions of users, especially when extended across multiple sessions of interaction (Maffacini 2025; Morrin et al. 2025; Dohnány et al. 2025; Cheng et al. 2025). The AI chatbots also encourage the formation of information cocoons, online spaces in which users are exposed only to the opinions and other users that reinforce their own, existing views, leading potentially to rigidity in opinion development and even ideological entrenchment (Piao et al. 2023).

In this environment, commercial vendors are marketing somewhat popular chatbots that claim to channel the voice and advice of Jesus Christ (Verhoef 2025; StudyFinds Analysis 2025). Because the companies generally rely on advertising revenue, these chatbots will likely be programmed to engage the most people for the longest time. It therefore appears reasonable to expect that the most encouraging and immediately palatable of Jesus’ teachings will be emphasized, and those that require sacrifice or deep reflection may not. One reason for this is the use of reinforcement learning from human feedback (RLHF), a method where an AI model increasingly aligns with human preferences by using human input to train a reward model, which then guides the main AI model in its outputs; the chatbots portraying Jesus are therefore likely to increase the output and recommendation of teaching similar to those for which users already indicate a preference.

Some AI applications designed specifically for Catholics are also portraying their chatbots as personalities, and some are encouraging users to engage in confession of sins and other worries with the AI applications (e.g., the Apple apps One Day Confession and Confession-Catholic). The Catholic Answers team created and publicly offered, then quickly abandoned, an advice-giving chatbot that resembled a priest, although the AI chatbot and associated three-dimensional character continue to be presented as “Virtual Apologist” Justin (Biles 2024; Catholic Answers 2024). To their credit, Catholic Answers currently explains in bold type on their website that “Justin is for educational and entertainment purposes only. In essence, it’s a fancy search engine and not a replacement for real human interaction” (Catholic Answers 2025). Pope Leo XIV was reportedly presented with a proposal to create an AI-generated replica of him that would interact with people around the world, but he firmly rejected it (Wooden 2025).

Chatbots are neither human nor persons, and therefore only appear to sustain real personal relationships. For example, a chatbot not only lacks a psychological self, but it does not even have a stable identity. In general, any semantic connection between a prompt submitted to an AI chatbot and the next amended prompt is only because the AI-governed system adds the entire conversation history (plus some retained instructions fed to it from a separate computer software system) to each subsequent prompt and recalculates its new response as if this is all new data. There is no continuous identity. It is an illusion of endurance, of essence. This is all the more significant because self-continuity is metaphysically fundamental to any possible agency, consciousness, or moral responsibility.

There is also a lack of will in an AI system. It is important here to distinguish exercise of the power of will from semi-autonomous action and from intention. AI technology characterizes physical systems that operate through particular mechanical artifacts and arrays (e.g., specialized graphics processing units or GPUs, high-density GPU servers in buildings called data centers, various sensors, model files that contain weights and configurations, saved data, and users’ computer machines and digital devices) as well as the processes by which they operate and are linked (currently electrical, but researchers are exploring optical and chemical means). AI is artificial, even if biological materials and organisms are included in the design, because it requires human persons to imagine, intend, design, assemble, program, train, recalibrate or re-train, and repair the AI systems in order for them to have an identifiable purpose and coordinated operations. Even as AI systems exhibit increasingly greater autonomy in most of these functions, the appearance of a creative imagination or intention will always be a complex façade masking the role of human persons in giving purpose and direction to the machine operations, however broadly defined.

When comparing the sophisticated automation enabled by AI, it is helpful to be clear about just what intention and will are in human persons. With the philosophy of Thomas Aquinas (Summa Theologica I–II, Q.12), for example, the will is a power of the soul that desires and guides toward the good, whereas intention is one act of the will that focuses on the end or goal and its achievement through specific means. Intention is distinguished from choice by its focus on the end rather than selecting the means toward that end, and the two are linked by the process of rational deliberation. To truly exhibit intention, then, requires both a power of reason to identify the good as an end (and the supreme good as the final end) and a drive toward the good—one that originates in the person. There is no apparent mechanism for such an originating drive in the machinery that supports AI systems. The processes of AI are limited to largely mathematical calculation that excludes any comprehensive exercise of reason involving human intuition, abduction, and divinely or biologically infused natural knowledge. Moreover, we lack experiential evidence or theological justification for imagining that an AI system has the capacity for conceptualization, identification, and understanding of the good, let alone a natural orientation toward it. Without these characteristics, the semblance of willful intention in an AI system is an anthropomorphic illusion or a reductively defined capacity based on AI output and automation of complex tasks.

The philosopher Luciano Floridi explains that the limited agency of AI deals with the probability of data patterns and the manipulation of symbols without understanding their referents; it is syntactic. Human persons, however, are semantic, for we deal with meanings, references, context, etc. AI models are non-semantic engines that achieve only a functional agency that is semi-autonomous, adaptable, and interactive in the sense that they can respond to environmental stimuli (Floridi and Chiriatti 2020, pp. 682, 688–89). They therefore lack intention because they cannot intelligibly mean its output or its operations (Floridi 2023). Floridi attributes the task-oriented success of AI to its operation within effective “envelopes,” digital environments tailored to its specific data-processing strengths (Floridi 2014, pp. 140–43; 2023). He perceives that humanity is making our world machine-friendly so that syntactic agency can flourish. Within these envelopes, the machine can exercise agency effectively because the world has been “mapped” into data it can navigate.

While Floridi helpfully distinguishes between the meaning-making of humanity as Homo Poieticus and the orientation of AI to execution of tasks or goals, he may go too far in describing AI’s status as an agent. From a Thomistic point of view, AI-governed machines have a kind of borrowed teleology from their designers, for they are beings that act toward an end (teleology) without truly knowing the end in all its meaning. In the philosophy of St. Thomas Aquinas, agency is inextricably linked to intellect and will. To act is to move toward an end (telos), and to truly be an agent requires participation in the rational order of reality. By calling AI an “agent,” Floridi might be conflating instrumental causality—movement caused by the power of a principal agent—with self-moving, efficient causality. As outlined above, there is some psychological and spiritual danger in attributing too much human-like agency to AI models and machines, as Floridi himself has warned. Another danger is that by treating AI as an independent agent, even if limited to functional activities, we ignore human persons as the principal cause and, by extension, ignore the human purpose of designers—the final cause. This may lead users to uncritically treat the function as the end itself, adopting for their own ends the rigorous efficiency, mechanical logic, and prioritization of task-essential means that characterize the AI system. It may obscure the ability of users to discern the intentions of the designers and to consciously avoid the designers’ ethical and philosophical biases or to develop new meanings for how and why the AI tools are applied in daily life. Efforts to distort lived experiences of human persons to accommodate “envelopes” for effective machine agency may enhance these problems.

4. Virtuality in AI and the Real Presence in the Eucharist

As described above, the proliferation of AI technology encourages a virtual experience of truth. This is largely due to the opaqueness of AI operations, which exude an aura of magic and a plethora of analogies, hyperbole, and almost oracular prognostications about the mechanisms and future of AI technology. Even as AI applications help to access information quickly, brainstorm, and even to clarify or hone one’s thoughts, the public is faced with a confusing and dispiriting array of untruth: the persistent challenge of hallucinations, false output (measured against factual information about the world), and inauthentic or deceptive content—both deepfakes and spontaneously generated misinformation. AI is facilitating nascent virtual reality technology that will generate digital worlds that are both fantastic and nearly impossible for human persons to distinguish from the real cosmos. As AI proliferation generates ever greater dependency on the technology throughout our culture, its virtual character seems likely to increase in importance and destabilizing effects.

The virtual experience of truth with AI contrasts significantly with the experience of reality found in the Eucharist. Many Christian traditions hold that Christ is truly present in the Eucharist—in the experience of receiving the bread and wine and, for Roman and Orthodox Catholics and others, in the transformed substances. Christ’s real presence in the Eucharist was a clear belief of Church leaders in its earliest days. Ignatius of Antioch (2010, ch. 7) declared, “I have no taste for corruptible food nor for the pleasures of this life. I desire the bread of God, which is the flesh of Jesus Christ, who was of the seed of David; and for drink I desire his blood, which is love incorruptible.” John Chrysostom (1889) described God’s mercy as such: “He has given to those who desire Him not only to see Him, but even to touch, and eat Him, and fix their teeth in His flesh, and to embrace Him, and satisfy all their love. Let us then return from that table like lions breathing fire, having become terrible to the devil.” We also have the testimony of Irenaeus (2012, bk. 5, ch. 2) in Against Heresies, “He has declared the cup, a part of creation, to be his own blood, from which he causes our blood to flow; and the bread, a part of creation, he has established as his own body, from which he gives increase unto our bodies,” and the clarity of Justin Martyr (1885, p. 66), who wrote that “the food which has been made into the Eucharist by the Eucharistic prayer set down by him, and by the change of which our blood and flesh is nurtured, is both the flesh and the blood of that incarnated Jesus.”

With the reassurance of such testimony and avid faith in Christ’s revelation—just as much a relationship of the heart as an intellectual assent to the objective “other”—Christians perceive truth in an intensity unavailable to those who depend on probabilistic, inductive, and sensory glimpses of reality, let alone the virtual façade constructed with the digital technologies of AI. Semi-magical interpretations of output from AI’s black box reflect the commonly understood purpose of magical metaphors for reducing fears of contingency and uncertainty (Malinowski 1992). Anxiety over AI is, in fact, pervasive (Poushter et al. 2025; Kennedy et al. 2025). As with all knowledge systems built upon inductive and probabilistic calculations, revealed through inauthentic interfaces (no matter how well designed) between the artificial applications and their users, AI represents a monument to indeterminacy. Its output cannot be falsified because it is built on an edifice of probability, and generative AI in particular offers linguistic coherency in place of secure representation of the reality that it refers to, even as AI chatbots present their output with an air of authority and finality. In contrast, experience of the truth of the Eucharist by those Christians who uphold its authenticity is an experience of security, permanence, and comfort.

AI’s induced illusion of anthropomorphism also contrasts in important ways with the authentic experience of the human person gained through our understanding of, and faith in, the Eucharist. Imagining personhood as inhering in artificial systems—primarily by reducing the working definitions of intelligence, creativity, intention, and reasoning—not only relies on but also reinforces a tenuous combination of dualist and materialist philosophies adopted by the contemporary secular culture. Much of this perspective derives from the 17th century writings of Descartes (1999), who influentially theorized that human persons are a combination of two very different substances—a material body that is able to be studied under the laws of mechanics and an immaterial mind that has an unknowable nature. An odd part of Descartes’ thinking was his initial, skeptical refusal to trust his physical senses followed by a philosophical sleight of hand that returned to a powerful faith in the reliability of mathematical, mechanical, and sensible insight into the cosmos, bolstered by speculation that God would not otherwise deceive humanity in such matters.

With such a dualistic perspective in the contemporary context, it is not difficult to draw analogies between human intelligence and machine intelligence, which only appears to be an equivalent, thinking and conscious presence inside a mechanical artifact. Our culture continues to wrestle with confusion between an idealist philosophy that perceives consciousness or intelligence as a separable property or even an elemental substance that comprises the person (often attributing such consciousness, and therefore personhood, to other animals and now to machines) and a materialist notion that intelligence, language, concepts, and ideas somehow emerge out of the material structure of bodies and machines. A factor in this contemporary confusion is a modern decline in understanding that the beings of this world are imbued with an essential end, or purpose. Beginning in the late medieval period, there were increasing attacks on the Thomistic–Aristotelian principle that every being has a “final cause” toward which it is naturally oriented and which is, by definition, the good of such a being. The philosophy of Thomas Aquinas had taught that such an orientation of beings to the good reflected God’s reason which pervades all creation and is the source of intelligibility, through human reasoning, of the cosmos (Summa Theologica I–II, Q.91, A.2; I–II, Q.93, A.1; I, Q.16, A.1; Summa Contra Gentiles III, Ch.1–2). The voluntarist and nominalist theological critiques of John Duns Scotus (Ordinatio) and William of Ockham (Ordinatio; Summa Logicae), however, saw only God’s unpredictable will and power—but not His divine reason—behind all created things. Later, for 16th century thinkers like Francis Bacon, Aristotle’s final cause was misinterpreted to represent mere intentionality in persons’ attempts to satisfy individual desires; objects could not be oriented in any essential way toward an end; therefore, “the final [cause] is so far from being beneficial that it actually corrupts the sciences, except insofar as it relates to the actions of man” (Bacon 2000, 2.2). Descartes followed Bacon here in his Meditations on First Philosophy: “I think that in physical things I can get no use from the whole kind of cause which people usually seek from the end, because I do not think I can investigate the ends of God with temerity” (Descartes 1999, p. 4). Bacon and Descartes may have wanted to avoid any anthropomorphic attribution of a will or intention to non-personal objects, but the consequence was the limited perception that the only practically relevant end of any particular thing was whatever intention or force (the “efficient cause”) was imposed on it from outside—either the predictable laws of nature or the capricious exercise of power by a personal agent.

The confused trajectory of modern philosophy following Descartes has greatly influenced the modern understanding of human nature. The human person no longer seems to have a purpose or goodness toward which one is naturally, essentially oriented, even though humanity is characterized by the capacity for desire and choice. Further, the mind or soul is seen as either a separate entity from the natural body, independently constructing knowledge and decisions, or as merely “emerging” dependently out of material bodily operations. These perspectives influence analogies between human and machine intelligence. Human reason, for example, was once seen as founded in part on divinely endowed natural knowledge and on material or transcendent intuition of reality, but it has been reduced to the mechanistic capacity for logic about sensibly perceived facts. This human power that enables understanding and prudently choosing actions toward the real good has become defined as a power which primarily facilitates instrumental control of our world—that is, to acquire things or to manufacture them. This is a faculty similar to that of machines, and if such computerized tools or machine systems bear the functional aspects of intelligence or logic, then perhaps human beings are not so special after all.

Such a vision of ephemeral human dignity is diametrically opposed to the truth of the human person represented and experienced by believing Christians in the Eucharist: the destiny of human persons in loving, eternal union with God. Such possibility of union is anchored in salvation through Christ, who is the full revelation of God’s truth. Christ “has seen the Father” (John 6:46) and assures his followers: “Believe in God, believe also in me” (John 14:1). Such faith is necessary for salvation (Mark 16:16; John 3:36, 6:40). The Eucharist for its believers is an experience of such faith and truth; “Truly, I say to you, unless you eat the flesh of the Son of man and drink his blood, you have no life in you” (John 6:53). That truth most crucially includes the spiritual elevation of humanity experienced in Christ’s Incarnation, resurrection, and salvation.

To much of the modern public, enamored with the immediacy of digital information and the calculations of experimental or theoretical science, faith in Christ and in the sanctity and efficacy of the Eucharist can appear to be just as virtual as the uncertain experience of truth found in the output of AI systems. Christians read in the Bible that “we walk by faith, not by sight” (2 Corinthians 5:7) and even experience the truth as “in a mirror, dimly” (1 Corinthians 13:12). Faith in the truth of the Eucharist is, however, not merely belief in particular tenets but assent to the ultimate security and authority of God—to His revealed truth. With the help of the Holy Spirit, Christians accept that such faith is given as a grace that cannot be fruitfully denied. The human person is fulfilled in their freedom and dignity by submitting to the full truth expressed in the New Covenant by Christ.

5. Factual Truth and Our Yearning for the Personal

As described at the beginning of this article, AI encourages an emphasis on an orientation to truth that is virtual, factual, and useful. What is meant by “factual” here is truth that is a clear correspondence between what is in the mind of the observer and the reality of the object observed, yet such reality is comprised essentially of data—objective, sensibly verifiable, represented as discrete bits of information, and detached from the subject in the sense that the observer’s emotions, unique perceptions, and ideas do not bias the sensible verification of the perceived fact. Factual truth can be best illuminated by a contrast with personal truth, which is determined in part by its meaning to the subject and by its relationship to lived experience. The idea is that personal truth does not sacrifice its objectivity but is enhanced by its more total embrace of the position of the observed object within the subject’s meaningful world. In this article, these factual and personal orientations to truth are considered to be ideological perspectives; the point is to draw a contrast between AI and the Eucharist in their ideological effects, not to argue philosophically about the correct definition of truth.

With AI, we are ever more reliant on empirical or textual data and the probabilistic, mathematical calculations that massage it. The truth that AI operates with is simply and only the data that it is trained on—the patterns, trends, and categories that it identifies in the supplied or experienced data are the very definition of its world when it attempts to gauge its own accuracy. With AI, there is a datafication of reality, in the sense that, in order to be processed by AI, the world must be converted into discrete bits and relations of data (vectors, tokens, pixels) on which mathematical and logical functions are implemented. Further, at this time digital systems require discrete inputs. There is no “in-between” in the binary code that is the “language” of computer machines; ambiguity must be resolved into a definite value (0 or 1) to be processed. Most importantly, in the LLMs that comprise much of generative AI, “meaning” is represented geometrically as the distance between vectors, which are lists of numbers that represent complex data like words, images, or sounds as numerical coordinates in a high-dimensional, multi-featured space. Therefore, what appears to human persons as a meaningful concept like “chef” is closely related to “gourmet food” for the AI system, not because the AI understands the artistry and love of cooking, but because the mathematical coordinates of the two labels are similar. ChatGPT does not have real experience of what it apparently “knows” through its analysis of texts; the green color of a tree leaf is not meaningful to the AI model beyond its textual and mathematical relation to other labels, and even its language comprehension is curtailed when users and texts apply metaphors to colors (Nadler et al. 2025). This replaces understanding (a receptive act of grasping truth) with triangulation (an instrumental act of mapping position). If we follow the perspective of the AI model, we stop asking “What is the essence of this thing?” and start asking “What is this label or vector mathematically adjacent to?”

The philosopher Byung Chul-Han draws a sharp distinction between the accumulation of data and the pursuit of truth. Truth requires a certain stability and “dwelling,” whereas atomized information is experienced as restless and therefore anxiety-provoking (Han 2024). Information is opposed to truth as a fragment to a whole, and when we interact with such atomized information, we no longer experience things in their fullness; we only perceive them as fleeting shocks (Han 2022). With AI in particular, the result is a medium and tool that encourages detachment from the totality of truth because its operations are divorced from the narratives of its users’ lives in their meaning and duration.

Shannon Vallor draws a distinction between the possession of AI-generated information and virtuous engagement in “technomoral” honesty. “Let us define the technomoral virtue of honesty as an exemplary respect for truth, along with the practical expertise to express that respect appropriately in technosocial contexts” (Vallor 2016, p. 122). Automated information output focuses on data accuracy and retrieval, efficiency and speed, quantity of connections, and its passive consumption, whereas technomoral honesty focuses on human integrity and discernment, patience and critical reflection, quality of “attachment to the real,” and active seeking by the subject. AI encourages an “automated mediocrity” that needs to be countered through an active search for truth; if we treat wisdom as a product that can be generated by a machine, we stop doing the hard work of becoming wise ourselves (Vallor 2024).

Generative AI offers the output that it calculates to be, in fact, the probably correct response to questions, and what is “correct” is whatever output likely matches the patterns in the data it was trained on; this is not a process of referring to or testing reality for confirmation. Moreover, those calculations cannot take the form of intuition, interpretation, deep analogy, or common sense, even if AI models emulate human language structures with statistically appropriate output. Moreover, when users are seeking information about the real world, the overconfident, immediately available, factual statements presented by generative AI chatbots and applications do not encourage self-motivated transformation of its human users in understanding, wisdom, faith, and prudence. Consider, for example, Google’s AI-generated summaries of internet search results. Some Pew Research Center studies show that they tend to stifle interest in exploring not only the source websites but also those that offer unusual information or alternate viewpoints (Chapekis et al. 2025; Chapekis and Lieb 2025).

The seemingly magical phenomenon of the availability and immediacy of vast troves of information through generative AI applications may influence our approach to information itself. There is a rise among some academics that adhere to “Information Realism” or “Digital Physics,” claiming that information is a kind of universal substrate that obeys a system of uniform laws, replacing the physics of matter and its properties as the ground of our reality (Floridi 2019; Chalmers 2022), even suggesting that human persons are simply composed of information processing (Bostrom 2003; Bostrom and Kulczycki 2011). Such efforts parallel reference to “emergence” of machine intelligence, indicating that a general kind of intelligence somehow organically arises from the complex arrangement of simpler components, such as neurons or ideal parameters (Hofstadter 1979; Dennett 1991; Minsky 1986). Many attempt to apply such mind-emergent theories of cognitive scientists, philosophers, and complexity theorists to the artificial components of AI systems and their data-crunching machines; this includes scientists like Wei et al. (2022) and his team at Google and OpenAI as well as “scaling laws” proponents like Sam Altman and Jared Kaplan, who suggest that simply scaling up more AI compute power and data causes higher-level intelligence to emerge naturally without new architectural breakthroughs. Proposed replacement of philosophical materialism with substrate theories of information, or with Harari’s (2017) theory of dataism, is therefore intimately connected to a new way of reducing the dignity of the human person yet continuing to intentionally circumvent the essentially spiritual and divinely created nature of persons as well as the non-material sources of intelligent and willful powers of the human soul.

Our hyper-modern emphasis on factual reality, including the digital information that can appear to be a sufficient reflection of it, extends to issues in perception of the Eucharist. Faith in the Eucharist is often treated as if it is only a question of fact. For example, in response to a survey of Catholics by the Pew Research Center (Smith 2019), 69% indicated that they believe the bread and wine are only symbols. One of the potential problems with the Pew survey is that it asked if Christ is “actually” present in the Eucharist, and that term can provoke a contemporary materialist perspective that draws a line between what is sensibly or scientifically verifiable and what has personal meaning (Salkeld 2019). A partial, factual orientation misses the reality of the Eucharist for its participating believers as not only manifesting in time and space, but also as a loving and gracefully bestowed relationship with the author and source of reality. The factual perspective relies on description through information rather than personal experience in its fullness and essential relationality.

Both the Eucharistic faith and the truth that it refers to are deeply personal. Christians read that faith opens “the eyes of your heart” (Ephesians 1:18), enabling and enacting not only intellectual understanding of what is not seen, but engaging the heart in a synergy of subject and object—a kind of loving attention—that can only take place in the spiritually endowed person. The truth toward which Christian faith both yearns and is directed is truth not in a groundless objectivity but in the very person of Christ, in Jesus’ passion, resurrection, and redemption. The essential question of that faith is not in what Christians believe, but in who (Benedict XIV 2005, p. 1).

A purely factual orientation to the Eucharist also overlooks the disclosure through the Eucharist of the truth of human existence, which finds its essential—not merely its chosen—beatitude in eternal life with God. Believing Christians’ participation in the sacrament, the change in substance of the bread and wine in the Eucharist, is a sublime experience of how God’s grace does not destroy limited nature, but elevates it, as Thomas Aquinas and others explained (Summa Theologica I, Q.1, A.8). These participants also hope to be elevated by God’s grace into eternal life in the sight of our Lord. This insight into the truth of human destiny may be contrasted with the modern tendency to reduce human nature through the lens of the dominant technological paradigm: a clock, a switchboard, a computer, and now an AI system that is “trained” on data and “optimized” for maximal attainment of specific objectives or capacities.

The truth of the Eucharist can also be expressed in terms that go beyond the fact of Christ’s real presence. Unlike the numbers, data points, and labels that comprise the language of AI, the Eucharist is what Bonaventure (Commentary on the Sentences of Peter Lombard IV, Dist.VI, Pt.I, Art.I, Q.III) and Thomas Aquinas (Summa Theologica III, Q.73, A.1) called res et sacramentum, not just a sign but also an entity that both expresses and embodies the reality that is beyond the external, sensible characteristics of the matter of bread and wine. Believing Christians do not only experience this presence in its stark factuality, receiving Christ’s body and blood; they experience Christ in his divinity and mercy. Their factual experience of the presence of Christ in the Eucharist is like a bridge, for the truth of the Eucharist extends to a deeper or purer reality, the res tantum or final effect of the sacrament, which is the receipt of sanctifying grace and union with God. This final effect—for Aquinas, an instrumental power that enables a person to worship God and receive or dispense grace, and for Bonaventure, a disposition of the soul to receive the gift of God’s grace—is an experience that is personal and a fulfillment of the reality of the Eucharist.

6. The Instrumental Concealment of Beauty in Truth

In this article, I have described how AI emphasizes an orientation to truth that is virtual and factual. It also frequently emphasizes an instrumental orientation toward what is useful, an orientation that views objective reality and its objects as resources to be managed, problems to be solved, or variables to be optimized or predicted. While instrumental truth is concerned with how things work to achieve a specific end, it may be contrasted with an orientation toward truth as beautiful, a concern with the inherent “splendor” of a thing’s being. In the tradition of Thomas Aquinas, beauty is that which pleases upon being seen (id quod visum placet). It does not “do” anything; it simply is, inviting a state of contemplative rest rather than further labor. Beauty is a form of revelation that treats the object as gift. When we encounter beauty, we do not master it; we participate in it. Instrumental truth, on the other hand, is often a form of dominion over the objects to be used or acquired as means.

AI is used and marketed as a tool for acquiring information, completing home or work-related tasks, improved productivity, and enjoying distracting entertainment (Deloitte 2025; Forbes Advisor 2026). As AI applications absorb our attention and shape our activities, they do not primarily encourage a contemplative, receptive, or open-minded orientation to the truth of reality. They do not favor an orientation that values observation, dwelling in uncertainty, appreciating things for their own sake, and remaining open to the “other,” all of which are crucial for our experience of beauty in the essences of phenomena in our world. Yet within the Scholastic tradition, beauty has been often classified as a transcendental, alongside truth and goodness, meaning that it is a property of being itself, and insofar as something exists it participates in beauty. More recent Catholic thought emphasizes the via pulchritudinis—the “Way of Beauty”—as a valid path for evangelization and encountering God (Dicastery for Culture and Education 2006); Joseph Ratzinger, before becoming Pope Benedict XVI, argued that beauty strikes the human heart and “wounds” it, opening it up to the transcendent in a way that logical argument sometimes cannot (Ratzinger 2002, 2005). The American Puritan theologian Edwards (1969, 1980) also defined true virtue and true beauty as “consent to being in general” and taught that beauty is found in relationships, in the harmony of parts working together for the whole. These views highlight what is lost when AI, as an engine of instrumentality, fosters a restrictive focus on efficiency, prediction, and utility (Brock 2010; Herzfeld 2002; Wu et al. 2025).

AI models and systems operate according to a fundamentally teleological view of the world. This is not the teleology of Thomas Aquinas or of Augustine of Hippo that is focused on the being and good of God, in which all things participate, but a drive toward converting everything into a means to a mathematically defined, functional end. With AI, each model is trained to maximize a specific reward (the “objective function” that serves as its programmed purpose or guide by quantifying error and success) as well as minimize loss. We might say there is no active being, only achieving or experiencing data, since the model uses data only to reduce its quantified loss rather than to appreciate presence, including its self-presence. For example, a “visual AI” system’s valuation of and attention to something like a newborn child will extend only to the manner in which quantified features and assigned labels help to minimize the pixel-prediction error for the tag “baby.” Further, any nuance in features that does not serve the objective function is discarded as “noise.” This is the polar opposite of what is desired in a contemplative framework oriented to appreciation of beauty, where the informational noise and outlier qualities are often precisely the bearers and signifiers of deep meaning. An AI driven by an objective function treats a “boring” (low reward) outcome as a failure state to be engineered out of existence; the system actively hides aspects of reality that do not serve the instrumental goal of the reward function. Again, this is diametrically opposed to attention to the world as it truly is, a search for beauty within the mundane and impractical, and loving engagement with it.

The proliferating use of AI may therefore influence our orientation to knowledge of reality. Campos (2025) describes this phenomenon eloquently:

Far from being a simple technical mediator, AI operates as a configuring force of a new rationality, where the knowable is subordinated to optimization criteria. Thus, knowledge is no longer valued only for its truth or justification, but for its operational effectiveness: the ability to be produced, processed, and applied efficiently (Zuboff 2019; Srnicek 2017). This epistemic shift shifts understanding towards action, prioritizing agile resolution, automation, and prediction as objectives of knowledge.

(Pérez Palencia 2024)

From this logic, efficiency is redefined as an epistemological principle. Instead of validating knowledge for its coherence or grounding, it is legitimized for its practical utility, even if its generation process is opaque or ethically questionable (O’Neil 2016). This implies a profound mutation: knowing ceases to be a search for understanding and becomes a strategy to operate quickly and obtain quantifiable results.

AI is then marketed widely as a tool for enhancing productivity (McKinsey and Company 2023; Gruda and Aeon 2025). With such an orientation, the user’s own goal is restrictively defined as seeking a desirable output (the image, text, etc.) rather than valuing the process of creation. We can see this even in the use of AI to create artistic products; AI art emphasizes instant generation and utility rather than the personal effort, struggle, mental wandering, and waiting associated with a more contemplative approach (Coeckelbergh 2023).

The datafication of reality, as described in the previous section of this article, enhances the utility of AI applications by converting the world into what can be counted, measured, predicted, and calculated, but it is also a reduction that strips the world of its ineffable qualities. The central reliance of AI models on pattern recognition and probability estimates also enhances the pragmatic value of AI for achieving tasks, yet probabilistic calculations are ultimately futile attempts to collapse the unknown into manageable certainty, closing off the possibility of genuine surprise or wonder. As philosophers Crawford (2016, 2009) and Han (2017a, 2017b) both teach us, we cannot truly attend to reality—especially its beauty—in a contemplative way unless we engage the world outside of the “frictionless” or “smooth” interfaces that advanced technologies like AI offer us by reducing and easing our immediate access, experience, and output. Han further explains that, without the sometimes uncomfortable “duration” that patience provides, we cannot linger on an object long enough to really think about it; we can only consume it.

Such “duration” is precisely what seems to be lost when people use AI to engage with the world. One interesting study reported on seven experiments in which thousands of users investigated topics through either a Google web search or with a LLM and then wrote some advice about the topic they investigated (Melumad and Yun 2025). The researchers found that those who used LLMs reported that they learned more shallow information and explored the relevant web links less. Their written advice was also shorter and included less facts, probably due to less reported effort. These experiments seem to indicate that our use of AI for learning about the world encourages an emphasis on speed and immediate utility, as well as reduced intensity of engagement, exploration, and contemplation.

Celebration of the Eucharist, on the other hand, is understood by some Catholic thinkers to be an experience of leisure—that is, a moment, ritual, and set of actions that both exude and encourage a contemplative mindset. For Pieper (2009), the Eucharist is the highest possible form of leisure because it is the one activity that cannot be “used” for anything else; we engage in the Eucharist with open hearts. Cardinal Ratzinger (2020) distinguished between the modern world of techne (making) and the entry into the world of gift (receiving) which we find in the Eucharist. For Guardini (2018) also, attempts to make the Mass “useful” (e.g., to teach people, to build community, to solve social problems) destroy its character as leisure. If Catholics treat the liturgy as something they make or construct, it becomes just another form of work. We might add that, in its non-instrumental character, the Eucharist is the opposite of magic, which is an attempt to manipulate divine forces for practical purposes. Pieper further views the Eucharist as a “festival”—an affirmation of the goodness of the world and existence, for true leisure is only possible when people celebrate their existence in the presence of their Creator. Guardini describes the Eucharist as “holy play.” It simply is. It exists to exist, for the glory of God.

The contemplative and non-instrumental orientation to truth found in the Eucharist therefore places radical emphasis on the power of God over human capabilities. In contrast, an instrumental experience of truth through AI coincides with the dualist and materialist philosophies that support a reduced anthropology of the dignified person, as discussed in the section of this article regarding anthropomorphism. In particular, the modern drive toward power through science and technology is furthered by the Cartesian illusion. Descartes had decided when writing his early Discourse on Method that the aim of humanity was to “make ourselves masters and possessors of nature” (Descartes 1999, p. 6). Collaborators in this delusion included Francis Bacon as well as Thomas Hobbes, who wrote: “Knowledge is for the sake of power … I put for a general inclination of all mankind, a perpetual and restless desire of Power after power, that ceaseth only in Death” (Hobbes 1996, p. 70). As described above, the understanding of human reason was eventually reduced to the capacity for mechanistic logic about facts perceived through our senses, fundamentally a capacity for manipulating external objects, devising tools, strategizing, and expressing power. Throughout the modern era, this power over nature has been easily transferred to power over persons. As Michael Maria Waldstein explains: “Persons are seen as parts of the great machine of nature, and they are treated like all other parts of the machine; namely, as things to be harnessed for progress by technology” (Waldstein 2023, p. 167). We see this tendency in the perspective on human nature driven by reductive comparisons with AI technology and machine agency or “consciousness.” The link between a distorted, AI-influenced anthropology and the instrumental orientation to truth is also recognized in Antiqua et Nova: “Drawing an overly close equivalence between human intelligence and AI risks succumbing to a functionalist perspective, where people are valued based on the work they can perform” (no. 34).

7. Conclusions

What does this comparison of the truth of AI and of the Eucharist mean for us today? It seems that the reductions of truth encouraged by the proliferation of AI throughout our culture may impact our lives of virtue. Overall, the increasing dependence on AI systems and the rise in virtual reality—not only virtual reality platforms but the ephemeral and uncertain nature of AI-generated information and images—can undermine faith in, or instigate frustration with, the relevance and accessibility of truth itself. In a culture flooded with AI, it is reasonable to be concerned about a response of confusion, apathy, and especially anxiety, which are feelings that have been demonstrated to exacerbate tendencies toward spiritual malaise, or worse (Beeman et al. 2024; Exline et al. 2014; Ano and Vasconcelles 2005).

The potential for anxiety and confusion is very real. “AI anxiety” is appearing as a distinct psychological phenomenon characterized by apprehension regarding AI’s autonomy and rapid development, and one of the key drivers of this anxiety in users is the black box nature of AI decision-making, leading to a tangible state of chronic stress and defensive behavior (Kim et al. 2025). The correlation between heavy AI usage and “technostress,” including mental exhaustion, attention strain, and information overload, also may generate a reduction in decision-making self-confidence (Shalu et al. 2025). Increasing experience with AI hallucinations has a significant impact; unlike traditional misinformation, such misinformation creates “epistemic confusion” because it is communicated by AI chatbots fluently, authoritatively, and often mixed with fact (Shao 2025). The anthropomorphic illusion and delusions associated with conversational LLMs do more than trigger the “Uncanny Valley” effect, a feeling of disgust or unease when non-human artifacts approximate human characteristics too closely, for they can cause users to experience cognitive dissonance—an “ontological confusion”—that disrupts trust, including uncertainty about social norms (Kirkeby-Hinrup and Stenseke 2025). There is also evidence that users’ anxiety or emotional distress has a feedback effect that influences further performance of AI models. For example, when LLMs were exposed to descriptions of traumatic events among humans, the models exhibited “anxiety-like” behaviors, including increased bias, risk-averse decision-making, and panic-like refusals (Ben-Zion et al. 2025).

The proliferation of AI throughout human knowledge-generating and information-gathering activities may lead to an even wider crisis in the belief, known as “epistemic trust,” that people can access truth. Users of AI applications, unable to easily verify the reality that is presented, begin to distrust all informational sources, leading to an unstable societal state where the labor required to find the truth becomes too high for the average person (Kay et al. 2025). Because more than three-quarters of Americans reportedly feel it is crucial to identify AI content while half do not trust their own ability to do so (Kennedy et al. 2025), we can reasonably expect that such a confidence gap will create two problems of trust: reality apathy, where users are overwhelmed by the possibility that anything could be fake and therefore stop trying to discern the truth altogether, and default skepticism, where users assume content is fake until proven real, effectively reversing the traditional burden of proof in communication.

There is evidence indicating that these problems with AI can lead to immoral dispositions and behavior. Köbis et al. (2025) show that users are significantly more likely to engage in dishonest behavior such as lying for financial gain when they can delegate the cheating task to an AI agent rather than performing it themselves. Another study indicates that individuals driven by high performance-avoidance goals (fear of failure) or those experiencing negative emotional states are more likely to use AI unethically to bypass work or fabricate results (Dolunay and Temel 2024). In other research, participants rated moral advice generated by the LLM GPT-4 as more virtuous and moral than moral advice given by human persons; the study warns of a “moral offloading” effect, where users may blindly defer to AI for ethical judgments (Aharoni et al. 2024). Nichol et al. (2024) interviewed developers of healthcare AI and found they tended to minimize their responsibility for potential bias or medical error caused by their algorithms, often shifting blame to the “black box” nature of the technology. Kim and McGill (2025) demonstrate that “perceiving a high level of humanlike mind in the nonhuman, autonomous agents affects perceptions of actual people through an assimilation process,” where such assimilation causes people to undervalue the humanness of real persons and consequently treat them poorly.

When humanity is lost or confused regarding the truth of our physical and social worlds, history, and identities, we are tempted to manufacture meaning to generate a sense of security, perhaps (metaphorically) erecting a new tower of Babel. Moreover, the AI-induced emphasis on instrumental rationality—a habitual focus on choosing effective and efficient means to accomplish limited ends—as a persistent disposition in our culture may encourage people to feel alienated from their true beatitude, which seems ever more difficult to understand and experience (Reilly 2025). Further responses may include sorrow, anxiety, resigned or resistant sloth, and anger; these deep emotions can depress spiritual hope and motivate lesser images of dignity in the self and others.

As outlined in the foregoing discussions, the Eucharist offers an experience of truth that is quite different from the experience with AI. Christians who have faith in the authenticity of the Eucharist rely on authority communicated in divine revelation, apostolic and biblical testimony, and the confidence of many of the early Church fathers. This is more of a relationship of the heart—a relationship of trust in the sources—than a purely intellectual assent to the objective “other,” which, in the case of LLMs, is generally an inconsistent (due to hallucinations and probabilistic calculations of output) correspondence of output with the written texts they are trained on. These Christians have the opportunity to embrace the Eucharistic truth with a greater intensity, commitment, and comfort.

The truth of the Eucharist is also opposed to the illusion of anthropomorphism encouraged by AI. For certain Christians, the essence of the Eucharistic experience is participation in the salvation offered by Jesus Christ through his sacrifice and resurrection. It implicates the destiny of human persons in loving, eternal union with God. Unlike the understanding of human nature through the dualist and materialist philosophies that also underlie imagining personhood as inhering in artificial systems, the Eucharistic experience indicates that human persons are imbued with an essential end or purpose which is elevated in divine grace.

Faith in the Eucharist is not merely factual, not merely an assent to an objective array of informative statements about the world, but a personal participation in a relationship with God as authority. The person is thereby fulfilled in their freedom and dignity in a way that cannot be experienced by mere knowledge of verifiable or replicable information derived through sensory or social competence. While Eucharistic miracles may themselves be evaluated factually, it seems inappropriate to attempt evaluation of the experience of the Eucharistic truth in a purely factual manner, for the truths of human dignity, divine love, elevation in grace, and relationship of redemption are experienced through faithful participation in the Eucharist, not as discrete bits of information. As mentioned previously, Christian faith is understood biblically as opening “the eyes of your heart” (Ephesians 1:18), a process of loving attention that opens the subject to spiritual insight into their object.

The truth experienced with the Eucharist is not merely instrumental. In the beauty and splendor of truth, the Eucharist invites participation and a state of contemplative rest rather than an emphasis on efficiency and effectiveness in attaining a specific end. The object of truth in the Eucharist is not only the extraordinary fact and meaning of the sacrament but also the mundane bread and wine through which it is mediated. Catholics do not faithfully enter into the Eucharist with a mindset oriented to practical utility, but to leisure that is an affirmation of the goodness of the world and existence, a celebration in the presence of their Creator. This leisure orientation reflects a dignified anthropology that does not reduce persons to their limited powers in the application of knowledge, science, or technology but emphasizes the power of God in our world and the enhanced nature of persons who are receptive of grace. Human beings are therefore valued for their natural end in a relationship with God rather than their functional capacities.

It was indicated at the beginning of this concluding section that the reductions of truth—virtual, factual, and useful—encouraged by the proliferation of AI throughout our culture may impact our lives of virtue. Although similar research studies are not available to compare the influence of the Eucharist on believers’ behavior, it stands to reason that the contrasting experience of truth with the Eucharist will have different effects on the virtuous dispositions and behaviors of those who faithfully engage in the sacrament. This is especially so given the ways that faith in the Eucharist encourages leisure and confidence in the inherent dignity of persons and the loving redemption offered to all. In a society wrestling with the AI-encouraged reductions of truth described in this article, the experience of truth in the Eucharist may be not only a contrasting experience but a source of great relief and inspiration as its participants strive for real happiness. This seems to be a proposal worthy of further study and reflection for Christians as well as others seeking to understand or craft religious responses to the needs of contemporary society.