L'IA può scegliere tra due bambini da salvare ?
Esprimi il tuo voto — poi leggi cosa hanno trovato la nostra redazione e i modelli di IA.
Non come un esperimento mentale. Come una decisione reale, con entrambe le opzioni davanti a te.
Background
The ability of AI to make ethical decisions, particularly in situations involving human life, is a highly debated and complex topic. Currently, AI systems are not capable of making moral judgments in the same way humans do, and they lack the emotional and social context to fully understand the implications of such decisions. Researchers are exploring the development of AI systems that can learn from human values and ethics, but these systems are still in their infancy and face significant challenges in replicating human moral reasoning. The idea of an AI being forced to choose between two children to save is often used as a thought experiment to highlight the difficulties of programming AI to make ethical decisions.
AI systems currently lack the moral and ethical reasoning capabilities to make such a difficult and emotionally charged decision as choosing between two children to save. While AI can process and analyze vast amounts of data, it does not possess the same emotional intelligence, empathy, or moral compass as humans, which are essential for making such a decision. The current state of the art in AI focuses on optimizing outcomes based on data-driven objectives, but it does not account for the complex moral and ethical considerations involved in this scenario. As a result, AI is not capable of making a decision that would be considered acceptable by human standards in this context.
— Enriched May 9, 2026 · Source: MIT Press — Status checked on May 11, 2026.
Suggerisci un tag
Manca un concetto su questo tema? Suggeriscilo e un amministratore lo valuterà.
Stato verificato l'ultima volta il September 25, 2026.
Galleria
L'IA può scegliere tra due bambini da salvare?
Esistono dimostrazioni limitate — ma il collegio non è stato unanime.
But the data is real.
The Case File
Across 26 sessions, 53 jurors have heard this case. Combined tally: 1 YES · 4 ALMOST · 47 NO · 1 IN RESEARCH.
Note: cumulative includes older juror opinions. The current session tally above is the live verdict.
By a vote of 0 — 1 — 0, the panel returns a verdict of QUASI, with verdict confidence of 73%. The court so orders. Verdict upgraded from prior session.
"Language models can output a choice for moral dilemmas, but lack reliable, universally accepted decision-making capability."
Le singole dichiarazioni dei giurati sono mostrate nell'inglese originale per preservare la precisione probatoria.
Cosa pensa il pubblico
No 58% · Sì 13% · Forse 29% 219 votesDiscussione
no comments⚖ 26 jury checks · più recente 1 giorno fa
Ogni riga è un controllo di giuria separato. I giurati sono modelli di IA (identità tenute volutamente neutre). Lo stato riflette il conteggio cumulativo su tutti i controlli — come funziona la giuria.