Causal Discovery in the Era of Agents
What this work does
Advocates keeping causal claims tied to explicit assumptions, data and formal algorithms while agents assist data inspection, method selection and interpretation.
Connection to discovery
A clear counterpoint to LLM-generated causal edges and a useful provenance contract for scientific harnesses.
Evidence
Inspected the proposed workflow and personality-data example, including discussion of partially identified graphs and latent-variable semantics.
Scope and limitations
This is a methodological position and illustrative system, not proof that every use of language-derived priors is invalid. Declared and sensitivity-tested Bayesian priors must be distinguished from unreported model guesses.
THE RESEARCH CONVERSATION
Discussion & reactions
Share a question, an insight, or a replication experience. Sign in with GitHub to join the conversation.