Whose Alignment? Comparing LLM Process Alignment Across Diverse Organizational Decision Contexts
The paper discusses the alignment of AI systems with organizational decision-making, emphasizing the complexity of this challenge. It introduces a method for measuring process alignment, which assesses how well an AI model weights information compared to an organization. The findings reveal that high process alignment is not always achievable or desirable, particularly in contested domains.
- ▪The study critiques the common framing of AI alignment as a single-target problem.
- ▪A decision-policy capturing method was used to measure process alignment in AI models.
- ▪Results showed that process alignment strongly predicts output accuracy in some contexts but not in others.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.25256 (cs) [Submitted on 24 May 2026] Title:Whose Alignment? Comparing LLM Process Alignment Across Diverse Organizational Decision Contexts Authors:Niklas Weller, Emilio Barkett View a PDF of the paper titled Whose Alignment? Comparing LLM Process Alignment Across Diverse Organizational Decision Contexts, by Niklas Weller and 1 other authors View PDF HTML (experimental) Abstract:Aligning AI systems with organizational decision-making is typically framed as a single-target problem: make the model behave like the organization. We argue this framing obscures a deeper pluralistic challenge.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
