Strategic AI-human collaboration for biology assessment design: a practical workflow

Area of Science:

• Biology Education
• Educational Technology
• Artificial Intelligence in Education

Background:

• Generative artificial intelligence (AI) tools are increasingly adopted by biology instructors for creating assessment rubrics.
• A significant gap exists in systematic workflows to ensure the scientific accuracy of AI-generated rubric criteria.
• This necessitates a structured approach for effective AI integration in biology assessments.

Purpose of the Study:

• To present a validated five-phase workflow for strategic AI-human collaboration in developing scientifically accurate assessment rubrics for undergraduate biology courses.
• To outline a process that integrates AI efficiency with essential human expertise for robust evaluation criteria.

Main Methods:

• A five-phase workflow was developed and tested across diverse undergraduate biology assessments (lab reports, genetics problems, discussions).
• Phases included: human-led learning outcome definition, AI-assisted criteria generation, human validation with biology-specific refinements, collaborative testing with sample work, and human-led implementation.
• The workflow incorporates decision rules, validation strategies, and practical examples for maintaining scientific rigor.

Main Results:

• The workflow effectively balances AI-driven efficiency in initial rubric generation with crucial human expertise in scientific reasoning and discipline-specific evaluation.
• Field testing across various biology assessments demonstrated the workflow's utility in maintaining scientific accuracy.
• The process ensures AI-generated criteria are refined and validated by subject matter experts.

Conclusions:

• The presented five-phase workflow provides a systematic and effective method for biology instructors to leverage generative AI for assessment rubric creation while upholding scientific integrity.
• Strategic AI-human collaboration is key to maximizing AI's benefits in education without compromising assessment quality.
• This approach supports the accurate and efficient development of assessment tools in biology education.