← Back to the overview

AI-Generated Protocol Assessment

Train students to identify practical, technical, and scientific flaws in experimental protocols through systematic evaluation.

Disciplinary ExpertiseLevel 2Complexity: AdvancedDuration: 30-60 min

Recommended for: Laboratory courses with emphasis on experimental technique

Procedure

  1. Students generate or receive an AI-created experimental protocol for a standard procedure relevant to the course (e.g., PCR, Western blot, chromatography, spectroscopic analysis, survey administration)
  2. Students conduct systematic protocol review, evaluating:
    • Missing controls: Are all necessary experimental controls explicitly specified?
    • Unrealistic steps: Are any procedures impractical, impossible to execute, or physically unfeasible?
    • Incorrect reagent specifications: Are concentrations, volumes, purity grades, or preparation methods incorrect or inadequate?
    • Timing errors: Are incubation times, reaction durations, or procedural sequences incorrect or suboptimal?
    • Safety omissions: Are necessary safety precautions, protective equipment, or waste disposal procedures missing?
    • Equipment specifications: Are instruments, materials, or apparatus inadequately specified or incorrectly designated?
    • Data collection procedures: Are measurement methods, recording procedures, and quality control steps adequate?
    • Reproducibility: Is sufficient procedural detail provided for independent replication?
  3. Students annotate the protocol with specific corrections, additions, and explanatory justifications
  4. Where facilities and time permit, students attempt to execute either the original or corrected protocol, documenting any additional issues encountered

Extension options

  • Compare protocols for identical procedures generated by multiple AI platforms or scientific search engines
  • Evaluate which systems produce more reliable, complete, and technically accurate laboratory protocols
  • In laboratory settings, systematically test AI-generated protocols to empirically identify problems

Discussion questions

  • What categories of protocol errors were most common, most serious, or most likely to compromise experimental results?
  • How might following a flawed or incomplete AI-generated protocol affect experimental outcomes, data quality, or safety?
  • What verification and validation steps should researchers implement when considering AI-generated protocols?
  • How does the quality and reliability of AI-generated protocols compare to protocols developed by experienced researchers or published in peer-reviewed methods papers?

Assessment opportunity

Students produce an annotated, corrected protocol with a detailed error analysis explaining each identified problem and its potential consequences.