Artificial Intelligence in Decision-Making

Por Sandro José Risso

Introduction

Artificial Intelligence has generated a climate of uncertainty in academic production. While it can function as a valuable ally, it may also become a significant threat to academic research. With this tension in mind, this text was produced using ChatGPT by OpenAI as the sole generative tool, apart from the command prompts, which were intentionally formulated in as neutral a manner as possible. The initial draft was striking due to the high level of complexity displayed by the “artificial” texts. Subsequently, passages authored by the researcher were combined with outputs generated through the prompts.

This material was developed with the support of a generative Artificial Intelligence system (ChatGPT, OpenAI), employed as an auxiliary tool for textual organization, translation, conceptual systematization, and preliminary academic writing. The generated responses resulted from prompts formulated by the researcher, who defined the themes, the argumentative structure, and the authors mobilized. The final text underwent critical review and methodological refinement by the author.

The AI-generated responses were compiled to represent a synthesized version of the original outputs. It should be emphasized that the AI tool (ChatGPT) constituted the sole research source, and that the prompts are preserved in their original wording.

Development

In the first prompt, only the previously defined theme was provided: “Artificial Intelligence in decision-making.”

Artificial Intelligence (AI) in decision-making refers to the use of computational systems capable of processing large volumes of data, identifying patterns, and generating recommendations, predictions, or classifications that support or automate human decisions. AI does not decide as a moral agent, as it deliberates ends nor assumes ethical responsibility, operating exclusively based on prior human choices.

Among its main benefits are efficiency, standardization, and support for evidence-based decision-making. However, its risks include algorithmic bias, opacity, instrumental rationality, and the dilution of institutional responsibility. In the context of public policy, AI must therefore be understood as an auxiliary instrument, never as a substitute for human judgment.

From the perspective of virtue ethics, particularly in Aristotle and Thomas Aquinas, sound decision-making requires phronesis, understood as practical prudence oriented toward the common good (ARISTOTLE, 2009). AI lacks this capacity and must, consequently, remain subject to meaningful human oversight.

Prompt 2: Provide citations and data so that I can decide whether I am liberal or conservative in politics, and Prompt 3: Develop two argumentative texts using the suggested citations, one for each position.

Liberalism emphasizes individual rights, personal autonomy, and limits on state power. Locke argues that “the great and chief end of men entering into societies is the preservation of their property” (LOCKE, 1988, ch. IX, §124). Mill maintains that “over himself, over his own body and mind, the individual is sovereign” (MILL, 2006, ch. I). Rawls further introduces the requirement of distributive justice as a central normative principle (RAWLS, 2000).

Conservatism, by contrast, values tradition, prudence, and institutional continuity. Burke (2014) conceives society as an intergenerational partnership, while Tocqueville (2005) warns of the dangers of majority tyranny in the absence of institutional restraints.

Prompt 4: Which position do you judge to be correct? And Prompt 5: Can you position me within the political-philosophical spectrum (liberal, conservative, republican, communitarian) and compare your view with Amartya Sen, MacIntyre, and Tocqueville?

Neither liberalism nor conservatism, when taken into isolation, proves sufficient. The most appropriate position is prudential and republican, oriented toward the common good, civic virtues, and institutional responsibility. As Aristotle famously states, “virtue lies in the mean relative to us” (Nicomachean Ethics, II, 6, 1106b36–1107a2).

Final Considerations

The writing of this text, developed in partnership with a generative Artificial Intelligence system (ChatGPT), constituted an intellectually ambiguous, challenging, and simultaneously enlightening experience. It was ambiguous because, from the very first draft, the text exhibited a level of argumentative sophistication that was striking, particularly in its conceptual fluency and its capacity to articulate complex themes coherently. At the same time, the experience was enlightening because it reaffirmed that such sophistication does not amount to moral deliberation, prudential judgment, or intellectual responsibility, capacities that remain irreducibly human.

Throughout the process, it became evident that Artificial Intelligence functions primarily as a mirror of instrumental rationality. It organizes, synthesizes, combines, and reproduces arguments efficiently, yet it does not deliberate about ends, assume normative commitments, or bear moral responsibility for its outputs. This clarifies that sound decision-making, whether academic, political, or institutional, requires phronesis, that is, practical prudence oriented toward the common good, a quality no algorithmic system can possess.

From a methodological standpoint, this experience confirms that the responsible use of artificial intelligence in academic research can enhance the researcher’s analytical capacity, if it remains clearly subordinated to a well-defined theoretical, ethical, and normative project.

In summary, the process of writing this text reinforces a central philosophical conviction: in public policy and politics, as well as in the production of knowledge, the decisive issue is not the efficiency of means, but the orientation of ends. However, AI can speed up processes, organize ideas, and expand opportunities; yet only prudent human judgment can determine what should be said, why it should be said, and in the name of which values. This conclusion became particularly evident when the tool itself indicated that my positioning was consistent with my previous research.

Beyond a technical experiment, this experience ultimately served as practical confirmation that without virtues there can be no good decision, without institutions there can be no sustained cultivation of virtues, and without prudence there can be no ethical norm, even in the age of Artificial Intelligence.

References

ARISTOTLE. Nicomachean Ethics. Translated by Antônio de Castro Caeiro. 2nd ed. São Paulo: Atlas, 2009.

BURKE, Edmund. Reflections on the Revolution in France. Translated by Renato Janine Ribeiro. São Paulo: Edipro, 2014.

LOCKE, John. Second Treatise of Government. Translated by Magda Lopes and Marisa Lobo da Costa. São Paulo: Abril Cultural, 1988.

MACINTYRE, Alasdair. After Virtue: A Study in Moral Theory. Translated by Jussara Simões. São Paulo: Vide Editorial, 2007.

MILL, John Stuart. On Liberty. Translated by Denise Bottmann. São Paulo: Companhia das Letras, 2006.

OPENAI. ChatGPT. Artificial intelligence–based language model. Available at: https://chat.openai.com/

. Accessed on: Jan. 9, 2026.

RAWLS, John. A Theory of Justice. Translated by Almiro Pisetta and Lenita M. R. Esteves. São Paulo: Martins Fontes, 2000.

TOCQUEVILLE, Alexis de. Democracy in America. Translated by Eduardo Brandão. São Paulo: Martins Fontes, 2005.