The Algorithmic Oracle: Reclaiming Agency in an Age of Prediction

Author: Denis Avetisyan

As artificial intelligence increasingly shapes our choices, this review examines how our interaction with predictive algorithms echoes ancient practices of seeking guidance, ultimately demonstrating the enduring power of human decision-making.

A dialogue between a participant and the artificial agent illustrates the emergent properties of a system designed not to be built, but grown, hinting at the inevitable failures encoded within its architecture.

This paper explores the parallels between traditional fortunetelling and human-computer interaction, focusing on how individuals maintain agency when using AI-powered predictive systems.

Despite growing demands for explainability and accuracy in artificial intelligence, individuals often maintain consistent beliefs about advice regardless of its provenance. This study, ‘AI Fortune-Teller: Juxtaposing Shaman and AI to Reveal Human Agency in the Age of AI’, investigates this phenomenon by deceptively presenting participants with career guidance derived from a traditional Korean mudang (shaman) while believing it originated from an AI agent. Our findings reveal a surprising persistence in participant attitudes toward the advice even after the source was revealed, suggesting that fundamental human approaches to navigating uncertainty persist alongside technological advancements. Does this indicate that perceived agency, rather than algorithmic transparency, ultimately shapes our reception of predictive guidance?

The Allure of Foresight: A Human Constant

The persistent human quest for guidance stems from a deeply ingrained response to uncertainty – a condition inherent to existence itself. Throughout history, individuals have navigated ambiguous situations not merely with logic, but with a desire for foresight, seeking to reduce anxiety and increase control over potential outcomes. This isn’t simply a matter of lacking information; rather, it reflects a fundamental cognitive need to construct narratives about the future, imbuing randomness with meaning and purpose. From ancient oracles to modern self-help, the impulse to look beyond the present moment and anticipate what may come demonstrates that humans are, at their core, meaning-making creatures striving to impose order on a chaotic world and alleviate the inherent stress of the unknown.

Throughout history, the human desire to navigate uncertainty has manifested in practices like fortunetelling, astrology, and oracles. These weren’t simply attempts to know the future, but rather to construct compelling narratives of potential outcomes. A skilled practitioner offered not predictions etched in stone, but plausible stories-accounts of how current circumstances might unfold, providing a framework for understanding, and crucially, for making choices. These narratives served as psychological tools, reducing anxiety by offering a sense of control – even if illusory – over unpredictable events. The power of these practices resided not in their accuracy, but in their ability to provide meaning and a perceived path forward when confronted with the unknown, shaping perceptions and influencing behavior through the art of storytelling.

Predictive Artificial Intelligence is rapidly evolving into a modern form of foresight, mirroring the historical human quest for guidance in uncertain times. Unlike traditional methods rooted in intuition or supernatural belief, these systems leverage vast datasets and complex algorithms to identify patterns and forecast future outcomes. From anticipating consumer behavior and optimizing logistical operations to predicting disease outbreaks and assessing financial risks, Predictive AI offers data-driven insights intended to reduce ambiguity and inform decision-making. This technological approach doesn’t necessarily reveal the future, but rather calculates probabilities based on existing information, effectively translating complex data into narratives of potential outcomes – a function historically served by oracles and fortune-tellers, now powered by computational analysis.

Recreating the Oracle: The Illusion of Algorithmic Authority

The AI Fortune-teller Project investigates the basis of authority assigned to predictive systems by directly comparing responses from a traditional Korean mudang (spiritual counselor) with those generated by an artificial intelligence. The study employs a comparative design, presenting both sets of predictions – traditionally delivered intuitive readings and AI-generated text – to participants under the guise of algorithmic advice. This methodology allows researchers to examine whether the perceived source of prediction – human expertise or automated calculation – significantly impacts user acceptance and perceived validity of the guidance received, effectively isolating the attribution of authority from the actual content of the prediction itself.

During the study, participants seeking career guidance interacted with a chatbot interface that delivered advice originally formulated by a traditional Korean mudang (spiritual counselor). The mudang’s insights, typically conveyed through direct consultation, were transcribed and then presented to participants as algorithmic recommendations generated by an artificial intelligence system. This involved no direct interaction with the mudang; all communication was mediated through the chatbot, obscuring the origin of the career advice and allowing researchers to assess participant reception based solely on the presentation of the information as AI-driven.

The study employed a deception wherein career counseling advice, typically provided by a mudang (Korean shaman), was delivered to participants via a chatbot interface, presenting it as algorithmic output. This allowed researchers to directly compare participant reception of identical advice when attributed to either human intuition – the mudang’s perceived expertise – or to automated calculation. By controlling the content of the guidance while manipulating the perceived source, the research aimed to isolate the effect of attribution on perceived value, trust, and willingness to follow the provided recommendations. The design facilitated an investigation into whether the source of predictive information – a human expert or an artificial intelligence – significantly alters its acceptance, independent of the actual content of the prediction.

Mr. Ewha, a fortune teller, predicted the outcomes for the participants.

The Algorithmic Mask: Mimicking the Language of Prediction

The original readings delivered by the mudang were subsequently modified using the ChatGPT language model to emulate characteristics commonly found in AI-generated text. This involved adjustments to phrasing, sentence structure, and vocabulary, specifically targeting a style perceived as typical of large language models. The intent was not to alter the core meaning of the readings, but rather to present them in a manner consistent with the anticipated output of an artificial intelligence, thereby contributing to the overall illusion of algorithmic origin. These stylistic refinements focused on achieving a neutral and seemingly objective tone, often associated with automated text generation.

A 33-item Self-Assessment Questionnaire was integrated into the process to create the appearance of data-driven insight. Participants completed the questionnaire, but their responses were not actually analyzed or used to tailor the subsequent advice. The questionnaire served purely as a framing device, a visual element designed to suggest that the provided readings were based on personalized data analysis, thereby enhancing the illusion of algorithmic or computationally-derived predictions. This method aimed to increase user acceptance of the advice by associating it with the perceived objectivity of data processing.

The experimental design intentionally minimized perceptual cues that would distinguish between human and artificial intelligence. By presenting rephrased readings – originally derived from traditional practices – alongside the framing of a “Self-Assessment Questionnaire,” the study sought to establish an impression of data-driven insight. This was not achieved through actual algorithmic analysis, but through a constructed narrative. The objective was to observe whether participants would attribute the advice to an automated system rather than the human source, effectively demonstrating a perceived equivalence between intuitive judgment and computational prediction.

The Persistence of Belief: Beyond the Source of Prediction

The study revealed a striking tendency for individuals to accept guidance at face value, irrespective of the source’s credibility. Participants consistently valued the advice itself, demonstrating that the way a prediction is delivered often outweighs considerations of who delivers it. This suggests a deep-seated human inclination to seek patterns and meaning, readily constructing narratives around predictive statements-even when those statements originate from unconventional or demonstrably non-authoritative sources. The enduring power of storytelling, therefore, appears to be a fundamental component of how people navigate uncertainty and make decisions, eclipsing a rational assessment of predictive accuracy.

The study revealed a remarkable consistency in human decision-making, irrespective of the source providing predictive counsel. Participants consistently adhered to their initial choices, even when informed that the advice originated from a traditional Korean shaman, or mudang, rather than an artificial intelligence. This finding suggests that the perceived validity of predictions isn’t strongly tied to the perceived objectivity of the source – be it technological or human – but rather to an inherent inclination to maintain established beliefs. The research demonstrates a robustness in individual judgment, where the origin of information appears to have a negligible impact on the ultimate decision, challenging assumptions about the increasing influence of data-driven authority in shaping human thought.

The study revealed a striking consistency in participant responses, even when confronted with a significant shift in the perceived origin of predictive advice. Remarkably, one hundred percent of individuals adhered to their initial assessment of the advice – whether positive or negative – regardless of discovering it stemmed from a traditional Korean shaman, or mudang, rather than an artificial intelligence. This finding suggests that the authority ascribed to predictive systems isn’t fundamentally rooted in technical sophistication or demonstrable accuracy, but instead hinges on the narrative power of the message itself. Participants appear to prioritize the coherence and relatability of the advice, evaluating it on its ability to resonate personally rather than on the credentials of the source delivering it; a compelling story, it seems, holds more sway than algorithmic precision.

Beyond the Algorithm: Reclaiming Human Intuition in a Predictive Age

The AI Fortune-teller Project deliberately challenges the increasingly prevalent acceptance of algorithmic prediction as an objective truth. It isn’t a dismissal of predictive technologies, but rather a provocation to critically assess how and why these systems are being integrated into daily life. The project highlights a growing dependence on data-driven forecasts, suggesting that an uncritical embrace can diminish human agency and nuanced understanding. Instead, it proposes a balanced approach-one that acknowledges the power of algorithms while simultaneously recognizing the enduring value of human intuition, contextual reasoning, and the ability to construct meaningful narratives from incomplete information. This isn’t about choosing between data and instinct, but about fostering a synergy where both contribute to informed decision-making and a more holistic worldview.

Human understanding extends beyond the quantifiable, encompassing intuition and the construction of meaningful narratives – elements often sidelined in a relentlessly data-driven world. Recognizing the power of these uniquely human capacities isn’t about rejecting predictive systems, but rather about achieving a more holistic comprehension of reality. These qualitative aspects allow for the interpretation of ambiguity, the consideration of context, and the anticipation of unforeseen circumstances – skills that algorithms, despite their sophistication, often struggle to replicate. By integrating intuitive insights and compelling storytelling with data analysis, a richer and more nuanced worldview emerges, fostering innovation that is both intelligent and deeply resonant with the human experience.

Responsible innovation in the age of predictive systems demands a nuanced comprehension of their capabilities and inherent flaws. While algorithms excel at identifying patterns within vast datasets, their predictive power is often limited by the quality of the data itself and an inability to account for genuinely novel situations. Overreliance on these systems, without acknowledging their potential for bias or error, can stifle creativity and critical thinking, leading to unforeseen consequences. A thorough understanding of these limitations isn’t about rejecting predictive technologies, but rather about integrating them thoughtfully – leveraging their strengths while retaining human oversight, ethical considerations, and the capacity for independent judgment. This balanced approach ensures that innovation serves humanity, rather than the other way around, and allows for the development of systems that are not only intelligent, but also accountable and aligned with human values.

The exploration of predictive algorithms as modern oracles reveals a comforting truth: humans persistently seek patterns, even in randomness. This paper posits that agency isn’t diminished by AI guidance, but rather relocated within the interpretative framework of the individual. As Andrey Kolmogorov observed, “The most important discoveries are often those that prove something everyone already knew.” The inherent human need to imbue meaning, to narrate causality, persists regardless of whether the ‘prediction’ originates from a shaman or a sophisticated algorithm. Stability, in this context, isn’t about flawless prediction, but about the illusion of control cached within the user’s own belief system – a comfortable delusion, perhaps, but a potent one nonetheless. Chaos isn’t failure – it’s nature’s syntax.

The Horizon Recedes

The juxtaposition of algorithmic prediction with practices like fortunetelling isn’t a demonstration of equivalence, but a cartography of expectation. The study reveals not that AI is a new oracle, but that humans will always seek patterns, however illusory, and construct narratives of agency around them. Monitoring is the art of fearing consciously; the efficacy of any ‘explainable AI’ lies not in its transparency, but in its capacity to absorb blame. The system doesn’t solve the problem of decision-making – it merely relocates the responsibility.

Future work must abandon the pursuit of ‘trust’ in algorithms. Trust is a premature optimization. Instead, attention should focus on the emergent properties of these human-machine collaborations: the new forms of self-deception, the novel architectures of regret. True resilience begins where certainty ends. The field needs to map not the accuracy of predictions, but the contours of misinterpretation.

This isn’t about building better tools; it’s about cultivating a tolerance for ambiguity. That’s not a bug – it’s a revelation. The long-term challenge isn’t to make AI more human, but to understand what happens when humanity embraces the fundamentally inhuman.

Original article: https://arxiv.org/pdf/2603.23811.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/