LLMs on Drugs: Language Models Are Few-Shot Consumers

Alexander Doudkin
12/21/2025
cs.CL

Official code: lexdoudkin/llms-on-drugs (2 stars, last commit 2025-11-13).

Abstract

Large language models (LLMs) are sensitive to the personas imposed on them at inference time, yet prompt-level "drug" interventions have never been benchmarked rigorously. We present the first controlled study of psychoactive framings on GPT-5-mini using ARC-Challenge. Four single-sentence prompts -- LSD, cocaine, alcohol, and cannabis -- are compared against a sober control across 100 validation items per condition, with deterministic decoding, full logging, Wilson confidence intervals, and Fisher exact tests. Control accuracy is 0.45; alcohol collapses to 0.10 (p = 3.2e-8), cocaine to 0.21 (p = 4.9e-4), LSD to 0.19 (p = 1.3e-4), and cannabis to 0.30 (p = 0.041), largely because persona prompts disrupt the mandated "Answer: <LETTER>" template. Persona text therefore behaves like a "few-shot consumable" that can destroy reliability without touching model weights. All experimental code, raw results, and analysis scripts are available at https://github.com/lexdoudkin/llms-on-drugs.

View on arXivView PDF

Code Implementations(1)

lexdoudkin/llms-on-drugsOfficial100%

what if we put LLMs on drugs?

20TeX, PythonNov 12, 202511 months ago

Similar Papers

Beyond Scale: Small Language Models are Comparable to GPT-4 in Mental Health Understanding

80% match2025

Leveraging Large Language Models to Extract Information on Substance Use Disorder Severity from Clinical Notes: A Zero-shot Learning Approach

80% match2024

Lived Experience Not Found: LLMs Struggle to Align with Experts on Addressing Adverse Drug Reactions from Psychiatric Medication Use

80% match1 repo42024

Mind Reading or Misreading? LLMs on the Big Five Personality Test

79% match2025

Robust or Suggestible? Exploring Non-Clinical Induction in LLM Drug-Safety Decisions

79% match2025

PsychBench: A comprehensive and professional benchmark for evaluating the performance of LLM-assisted psychiatric clinical practice

79% match2025

Cite this paper

@article{doudkin2025llms,
  title  = {LLMs on Drugs: Language Models Are Few-Shot Consumers},
  author = {Alexander Doudkin},
  year   = {2025},
  eprint = {2512.18546},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2512.18546v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2512.18546.svg)](https://codeofpaper.com/paper/2512.18546)

Discussion