Responsible AI in Education

• These guidelines ensure generative AI is used responsibly and ethically within education, benefiting students and educators
• The guidelines are aligned with Unesco’s “Guidance for generative AI in education and research” and MIT Sloan "Review on Responsible AI in Higher Education”

Data Privacy and Security

• Data handled by generative AI tools must be secure and ethical, protecting student data and complying with regulations
• Data security protects sensitive information, preventing unauthorized access and misuse, essential for responsible and ethical use in education

Transparency and Accountability

• Educational institutions must be transparent about AI tool use and decision-making processes
• The transparency includes explaining AI's role in learning and ensuring accountability for AI-generated content
• Institutions should define AI's goals and values collaboratively, reducing biases and misconceptions to align with ethical principles

Bias and Fairness

• AI tools require regular auditing to ensure outcomes are fair and equitable for all students
• Diverse training data and continuous monitoring of AI outputs are necessary to check for signs of bias

Large Language Models (LLMs)

• LLMs use statistical algorithms to analyze data and determine patterns so data input mimics data received
• If majority of data inputs relate to industry, language, demographic or time period the output will do so as well
• Generated content mirrors learned language patterns and received examples

LLM Chatbot Inherent Bias

• Users should be aware of potential bias in chatbot responses
• LLMs can provide inaccurate, misleading, and unethical information, so users must recognize that LLM output is based on human-written data and information
• LLMs share intellectual property without attribution so a "critical lens" should be used when AI is used

Academic Integrity

• Policies should prevent AI misuse for cheating or plagiarism
• Educating students of the ethical use of AI and implementing systems to detect and deter academic dishonesty is required

Human Oversight

• AI tools should complement, not replace, educators; human oversight maintains and enhances teaching and learning while preserving the educators' critical role

Inclusivity and Accessibility

• Ensure AI tools are accessible to all students, including those with disabilities
• User-friendly AI systems should provide the necessary support for effective use

Environmental Considerations

• Consider the environmental impact of AI tool use, including the energy consumption of training large models
• Sustainable practices can mitigate these effects

Continuous Evaluation and Improvement

• Regular evaluation is necessary to assess the effectiveness and impact of AI tools in education
• Feedback from students and educators along with making necessary adjustments improves performance and relevance