A Hidden Shift Is Not a Hidden Bias: What a Representation-Level Counterfactual Gap Certifies About an LLM Recommender, Why the Encoded-Versus-Expressed Studies Disagree, and the Placebo, Mediation and Transport Tests That Would Settle It

APA

Mahendrakar, P. (2026). A Hidden Shift Is Not a Hidden Bias: What a Representation-Level Counterfactual Gap Certifies About an LLM Recommender, Why the Encoded-Versus-Expressed Studies Disagree, and the Placebo, Mediation and Transport Tests That Would Settle It [Preprint]. Life of Research. https://doi.org/10.5281/zenodo.23437257

Copy APA

BibTeX

@misc{lor1152026,
  title        = {A Hidden Shift Is Not a Hidden Bias: What a Representation-Level Counterfactual Gap Certifies About an LLM Recommender, Why the Encoded-Versus-Expressed Studies Disagree, and the Placebo, Mediation and Transport Tests That Would Settle It},
  author       = {Pranay Mahendrakar},
  year         = {2026},
  publisher    = {Life of Research},
  doi          = {10.5281/zenodo.23437257},
  url          = {https://lifeofresearch.com/preprints/115},
  keywords     = {fairness evaluation, LLM-based recommendation, intrinsic and extrinsic bias, counterfactual fairness, representation probing, concept erasure, activation steering, causal mediation},
  note         = {LOR-00115}
}

Copy BibTeX More on the record page