EUVIMEDEUVIMED entdecken
Crossref

What Does Your AI Mean by “Flourishing”? The Case for Disclosing and Benchmarking the Values in AI Alignment

Alexandre Castonguay

JMIR Metascience and Research Integrity · 03.11.2026 · 1 · e106380

DOI: 10.2196/106380 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JMIR Metascience and Research Integrity, online erschienen am 03.11.2026. Verfasst von Alexandre Castonguay. Die Arbeit nennt 50 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Who could be against human flourishing? Every AI developer promises that its systems will serve humanity, and governments promise the same. It sounds like consensus. It is not: behind the shared promise lie very different pictures of a good life, rarely spelled out. Picture a patient facing a serious cancer diagnosis and asking an AI assistant for help. One assistant opens with survival curves. Another asks what the patient wants the coming months to hold. A third quietly weighs what the treatment will cost the hospital or the insurer. All 3 may pass the same capability and safety evaluations. None of those evaluations tells the patient whose idea of a good life is shaping the conversation. This Viewpoint argues that flourishing has become a third axis of AI evaluation, alongside capability and safety, and that the first step is not agreement on one definition of a good life. It is transparency: disclose, do not dictate. It proposes a Human Flourishing Statement—a short public declaration of what a developer means by flourishing, which dimensions it includes and leaves out, how it settles trade-offs and disagreements, how far the system is designed to influence its users, and what evidence shows that training moves the system toward those goals. A statement is not proof. It does not show that a system keeps its promises, and it does not license a system to steer anyone. What it does is turn a slogan into a commitment that others can test. Independent evaluations can then check whether behavior matches the declared target, and clinical value labels such as the proposed Values In the Model framework can show how those values play out in care. Health informatics knows this territory. It turned “trust the authors” into reporting standards and made patients' values part of clinical decisions. It is well placed to help developers, researchers, clinicians, and institutions make flourishing claims checkable, starting with voluntary statements and journal requests. Capability, safety, and a visibly declared conception of human flourishing should advance together, so that “beneficial AI” becomes a claim we can check rather than a slogan we are asked to trust.

Zitieren

APA

Castonguay, A. (2026). What Does Your AI Mean by “Flourishing”? The Case for Disclosing and Benchmarking the Values in AI Alignment. JMIR Metascience and Research Integrity, 1, e106380. https://doi.org/10.2196/106380

Vancouver

Castonguay A. What Does Your AI Mean by “Flourishing”? The Case for Disclosing and Benchmarking the Values in AI Alignment. JMIR Metascience and Research Integrity. 2026;1:e106380. doi:10.2196/106380

BibTeX
@article{Castonguay2026,
  author = {Castonguay, Alexandre},
  title = {What Does Your AI Mean by “Flourishing”? The Case for Disclosing and Benchmarking the Values in AI Alignment},
  journal = {JMIR Metascience and Research Integrity},
  year = {2026},
  volume = {1},
  pages = {e106380},
  doi = {10.2196/106380}
}
Kostenlos starten →