arXiv:2606.04978v1 Announce Type: new 
Abstract: LLMs can appear cautious in risk decision-making tasks, yet cautious-looking outputs do not necessarily indicate alignment with human decision-making mechanisms. We investigate this distinction using the St. Petersburg game as a controlled testbed, a classical paradox in which the expected payoff is infinite, yet humans typically report low, finite willingness to pay. We evaluate 28 LLMs with a structured prompt suite that includes the original game; controlled decision variants that perturb truncation, repeated play, numeric endowment, and occupational identity; a human-perspective prompt that asks models to reason as human decision makers; and paired comparisons between base models and their instruction-tuned counterparts. In the original game, most models generate finite bids, creating the appearance of human-like risk behavior. However, this outcome-level resemblance masks substantial mechanism-level differences. The controlled variants reveal that rather than maintaining human-like behavior seen in the original game, models often shift to conditionally and computationally rational behavior. Human-cue prompting and instruction tuning often lower bids and reduce some visible pathologies, but most mechanism-level response patterns remain largely unchanged. These findings show that behavioral alignment in risk decision-making can be surface-level: LLMs may produce human-like risk decisions without exhibiting human-consistent mechanisms. High-stakes evaluations of LLM decision-making should therefore move beyond outcome similarity and examine whether the alignment is supported by mechanism-level consistency.

أظهرت دراسة حديثة عمليات اتخاذ القرار لنماذج اللغة الكبيرة (LLMs) باستخدام لعبة سانت بطرسبرغ، وهي مفارقة كلاسيكية حيث تكون العوائد المتوقعة غير محدودة، لكن المشاركين البشر عادةً ما يعبرون عن استعداد محدود للدفع. قامت الدراسة بتقييم 28 نموذجًا، وكشفت أن العديد منها أنتج عروضًا محدودة تشبه السلوك البشري، لكن تم اكتشاف اختلافات كبيرة في آليات اتخاذ القرار.

Un estudio reciente examinó los procesos de toma de decisiones de los modelos de lenguaje grandes (LLMs) utilizando el juego de San Petersburgo, una paradoja clásica donde los pagos esperados son infinitos, pero los participantes humanos suelen expresar una disposición a pagar finita. La investigación evaluó 28 LLMs, revelando que, aunque muchos produjeron ofertas finitas similares al comportamiento humano, se descubrieron diferencias significativas en los mecanismos de toma de decisiones.

Une étude récente a examiné les processus de prise de décision des grands modèles de langage (LLMs) en utilisant le jeu de Saint-Pétersbourg, un paradoxe classique où les gains attendus sont infinis, mais où les participants humains expriment généralement une volonté de payer finie. La recherche a évalué 28 LLMs, révélant que bien que beaucoup aient produit des offres finies similaires au comportement humain, des différences significatives dans les mécanismes de prise de décision ont été mises en évidence.

A recent study examined the decision-making processes of large language models (LLMs) using the St. Petersburg game, a classic paradox where expected payoffs are infinite, yet human participants typically express a finite willingness to pay. The research evaluated 28 LLMs, revealing that while many produced finite bids similar to human behavior, significant differences in decision-making mechanisms were uncovered.