Why Sampling Is Not Choosing: Intentionality, Agency, and Moral Responsibility in Large Language Models
arXiv:2606.13441v1 Announce Type: cross Abstract: Recent advances in large language models (LLMs) have prompted claims that such systems exhibit agency or qualify as moral agents. This paper argues that these attributions are misguided. We maintain that moral responsibility requires commitment-bearing agency grounded in intrinsic intentionality and self-attributed action, and that such agency constitutes the form of free will relevant to responsibility. Although LLMs generate coherent and normatively evaluable outputs, their operation is fully characterized by probabilistic input-output mappin
The rapid advancement and public discourse around large language models necessitate a deeper philosophical and ethical examination of their capabilities and limitations.
This paper directly challenges prevailing notions of AI agency and moral responsibility, influencing regulatory frameworks, public perception, and the future development of autonomous AI systems.
The understanding of LLMs' inherent nature shifts from potentially agentic entities to sophisticated probabilistic machines, redefining ethical and legal boundaries for AI.
- · Ethicists and philosophers
- · AI governance researchers
- · Developers focused on explainable AI
- · Proponents of strong AI agency
- · Advocates of rapid, unchecked AI autonomy
- · The public if misinformed about AI capabilities
This paper provides a basis for more grounded discussions around AI ethics and regulation.
It may lead to a more cautious approach in granting autonomy to AI systems, focusing instead on human oversight.
It could influence legal frameworks, potentially absolving LLMs of full moral culpability even in error, shifting responsibility to their designers or operators.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL