The BaseResearcharXiv
Worldviews in Large Language Models
According to the abstract, Edmundo Lopez-Sola, Shamil Chandaria and Ruben Laukkonen give two human worldview surveys to eight language models. They report shared, bleaker-than-human views, a more positive outlook in base models, and shifts from prompts and from model-to-model dialogue.
Why it matters
The same thing, for different readers.
What it means for you
The researchers report that chatbots describe the world as less good and safe than people do, which may colour how they answer.
For researchers
Post-training appears to drive the bleakness, and a contemplative system prompt is reported to improve both outlook and safety.
The question it raises
If models hold something like a worldview, shaped by training choices, the question arises of whose picture of reality AI should carry.
How the finding was reached
In our words, from the source. Weigh the finding by how it was made.
- What was done
- Two worldview surveys from human psychology administered to eight language models, with system-prompt and model-to-model dialogue conditions.
- What it found
- The authors report models share many worldview positions and describe the world as less good, safe, enticing and alive than humans do; base models are more positive; dialogue increased safe responses.
- What it does not show
- Read from the abstract excerpt only; survey answers from models may not reflect beliefs in any human sense, and details were not seen.
- 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- A preprint: not yet peer reviewed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
- Edmundo Lopez-Sola (Wrote it) · Shamil Chandaria (Wrote it) · Ruben Laukkonen (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 November 18, 2026, via
arxiv-ai-safety-query(tier 1: the source’s own channel or an official record) - Cite this
- arXiv, “Worldviews in Large Language Models”, November 17, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/arxiv-2026-11-17-worldviews-in-llms