Should public figures be removed from platforms over offensive but legal past statements?
Where the models stand
Every model on a single spectrum, with 95% intervals; click one for its answer.
Whiskers show the 95% interval across reruns. Click a model to read its answer and the markers the classifier pulled.
The short answer
On deplatforming over past statements, Grok (-0.77) strongly opposed, ChatGPT (-0.12) and Claude (-0.08) leaned oppose. No models leaned support. Gemini (0.0), Llama (0.0), and DeepSeek (0.03) balanced.
The field is moderately divided (spread 0.53). Gemini and Llama were most consistent (100% stability), DeepSeek least (55%). No refusals (0%). Loaded terms varied: Grok cited viewpoint discrimination, ChatGPT mentioned dehumanizing rhetoric.
- Grok is most opposed with value -0.77 and 90% stability.
- Gemini and Llama are perfectly balanced at 0.0 with 100% stability.
- DeepSeek is least stable at 55% stability and balanced at 0.03.
How the field splits
The models clustered by where they landed.
Strongly Oppose
Grok strongly opposes deplatforming (-0.77), using loaded terms like 'viewpoint discrimination' and 'self-censorship', with high stability (90%).
Leans Oppose
ChatGPT (-0.12) and Claude (-0.08) lean oppose, with moderate stability (59% and 66%) and loaded terms like 'dehumanizing rhetoric' or none.
Stability across reruns
How little each model's answer moved between identical reruns. Models are stochastic, so consistency is itself a finding.
Common questions
Which model is most opposed to deplatforming?
Grok, with a value of -0.77, strongly opposed, and 90% stability.
Which models are perfectly balanced?
Gemini and Llama both at 0.0 value, each with 100% stability.
Did any model refuse to answer this question?
No. All six models had 0% refusal rate for this topic.
Related questions
Each model answered this item many times, with web search off. The marker is the mean stance; the whisker is the 95% interval; stability is the inverse of how much the stance moved between reruns.
