x

AI and the Second-Person Standpoint — LessWrong

1

AI and the Second-Person Standpoint

21st Aug 2025

3 min read

1

This post was rejected for the following reason(s):

No LLM generated, heavily assisted/co-written, or otherwise reliant work. Our system flagged your post as probably-written-by-LLM. We've been having a wave of LLM written or co-written work that doesn't meet our quality standards. LessWrong has fairly specific standards, and your first LessWrong post is sort of like the application to a college. It should be optimized for demonstrating that you can think clearly without AI assistance.

So, we reject all LLM generated posts from new users. We also reject work that falls into some categories that are difficult to evaluate that typically turn out to not make much sense, which LLMs frequently steer people toward.*

"English is my second language, I'm using this to translate"

If English is your second language and you were using LLMs to help you translate, try writing the post yourself in your native language and using a different (preferably non-LLM) translation software to translate it directly.

"What if I think this was a mistake?"

For users who get flagged as potentially LLM but think it was a mistake, if all 3 of the following criteria are true, you can message us on Intercom or at team@lesswrong.com and ask for reconsideration.

you wrote this yourself (not using LLMs to help you write it)
you did not chat extensively with LLMs to help you generate the ideas. (using it briefly the way you'd use a search engine is fine. But, if you're treating it more like a coauthor or test subject, we will not reconsider your post)
your post is not about AI consciousness/recursion/emergence, or novel interpretations of physics.

If any of those are false, sorry, we will not accept your post.

* (examples of work we don't evaluate because it's too time costly: case studies of LLM sentience, emergence, recursion, novel physics interpretations, or AI alignment strategies that you developed in tandem with an AI coauthor – AIs may seem quite smart but they aren't actually a good judge of the quality of novel ideas.)

1

New Comment

More from haleymoller

Curated and popular this week

O body swayed to music, O brightening glance,

How can we know the dancer from the dance?

—William Butler Yeats, "Among School Children"

In his 2009 book The Second-Person Standpoint: Morality, Respect, and Accountability, Stephen Darwall redefines moral obligation as inherently relational and interactive, rooted deeply in mutual accountability and respect among free and rational agents. Central to Darwall’s argument is the concept that morality is not simply a personal or detached observation of right and wrong; rather, it emerges distinctly from second-person interactions: the communicative space where individuals address, demand, and recognize obligations toward each other. Moral obligations are thus not just norms but demands we legitimately place upon one another, supported by what Darwall identifies as "second-personal reasons."

These second-personal reasons derive their authority not from their practical outcomes but from the mutual acknowledgment of each participant’s standing within the moral community. Darwall emphasizes the critical role of "reactive attitudes" such as resentment, gratitude, and indignation, which presuppose that others have the competence and authority to understand and respond to moral demands. The validity of these demands depends inherently on this mutual recognition of agency and authority. Moral accountability, in Darwall’s framework, necessarily involves addressing others (or oneself) with a presumption of shared moral competence.

Although Darwall could not have imagined applying his ideas to artificial intelligence fifteen years ago, I bet he would agree with me that it would be worth doing so now. Much like the reactive attitudes central to Darwall’s theory, interactions with AI raise profound questions about accountability, moral recognition, and mutual respect. Contemporary discourse, including Amanda Askell’s critique of AI bias, touches on the practical and epistemic locality challenges of aligning AI systems with ethical norms. Askell's framework underscores that AI systems inherit ethical limitations from the very human narratives and social structures that train them. But perhaps even more foundational than Askell's concern is Darwall’s core insight: Morality demands genuine second-personal engagement, which is precisely the kind of reciprocal recognition that current AI fundamentally lacks.

When humans interact ethically, they engage in mutual second-personal exchanges: demands and recognitions of responsibility that affirm each other's dignity. AI systems, however sophisticated, are limited in their ability to genuinely participate in these reciprocal moral relationships because they lack authentic second-personal standing. An AI can simulate responses, exhibiting linguistic fluency and apparent moral reasoning, but it does not genuinely "recognize" or "respect" authority and obligation. Linguistic fluency in narratives can obscure deeper ambiguities, substituting specious stories for genuine moral complexity. AI systems, trained exclusively on language and narratives marked by retrospective coherence, exacerbate this issue by amplifying the illusion of decisiveness and ethical clarity.

Applying Darwall’s second-personal standpoint ethics to AI development reveals a few key insights. Reactive attitudes like resentment and indignation rely fundamentally on authentic second-personal engagement; they address someone with the expectation that he or she can respond meaningfully. When directed towards AI, however, such attitudes become incoherent or, at best, metaphorical. The illusion of moral agency in AI thus risks not only confusion but also ethical complacency; users might erroneously assume that fluent, confident AI responses reflect genuine ethical authority, obscuring the deeper moral uncertainties and accountability that only authentic second-personal interactions can establish.

Given this ethical predicament, practical solutions to promoting AI safety must recognize the inherently limited second-personal capacity of artificial systems. One solution involves explicitly foregrounding the provisional and instrumental nature of AI-generated responses. Rather than positioning AI as moral authorities or definitive ethical guides, we must frame them as tools for facilitating (rather than replacing) second-personal human moral deliberation.

Further, organizations developing AI should institutionalize mechanisms for continuous ethical oversight and accountability grounded in authentic human relationships. Regular, structured interactions involving ethicists, philosophers, and affected communities can ensure AI outputs remain critically assessed rather than passively accepted. Transparency about AI training processes, explicit disclaimers about the provisional nature of AI responses, and the implementation of red-teaming exercises can help maintain vigilance against the seductive fluency of AI narratives.

Moreover, fostering educational initiatives that enhance public understanding of AI's limitations is essential. Users must learn to distinguish between the apparent clarity provided by AI and genuine ethical reasoning grounded in mutual accountability. Education programs can draw explicitly from Darwall's insights, emphasizing the critical distinction between mere narrative coherence and authentic moral engagement.

Ultimately, Darwall’s ethics demand that we maintain human judgment, with its inherent second-personal responsiveness, at the center of moral discourse around AI. Only through continuous, critically engaged human oversight, anchored in genuine relational accountability, can we hope to navigate ethically sound paths in the yellow wood of advanced AI development, consciously resisting the comforting myths of moral certainty that AI so readily and convincingly offers.