Model vs. User: Whose Values Prevail? Inference-Time Value Alignment in Multi-Turn Conversations

A preprint tests four steering methods in multi-turn conversations where user and assistant push toward the same or opposing values, and reports that the side matching the assistant's baseline bias often prevails.

Why it matters

The same thing, for different readers.

What it means for you

An AI's built-in leanings can win out over what a user wants over a longer conversation.

For researchers

Argues steering should be evaluated over full conversations and in both directions of a value trade-off.

The question it raises

Raises whose values prevail when a person and a model disagree.

How the finding was reached

In our words, from the source. Weigh the finding by how it was made.

What was done
Four prompt- and activation-based steering methods, three models, two value trade-offs, tracked with a linear probe.
What it found
When assistant and user pull in opposite directions, the side matching the assistant's baseline bias often prevails; early successes often fade or reverse.
What it does not show
Three models and two trade-offs; abstract only.
Status
A preprint: not yet peer reviewed

The record

Questions
What values should AI have, and could it have a mind of its own?
People
Enrico Liscio (Wrote it) · Anne Arzberger (Wrote it) · Aditya Shankar (Wrote it) · Pradeep K. Murukannaiah (Wrote it)
What we read
We have read only the description the publisher gives. No quotations until someone reads the source itself.
Found by
The Radar, run of October 15, 2026, via arxiv-ai-safety-query (tier 1: the source’s own channel or an official record)
Cite this
arXiv, “Model vs. User: Whose Values Prevail? Inference-Time Value Alignment in Multi-Turn Conversations”, October 14, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/arxiv-2026-10-14-model-vs-user-values
Our notesLight lane: written from title, author and abstract excerpt only; not fetched. Authors listed are those named in triage.