Skip to Main Content

AI in ECU Research: Transparency

Always disclose the use of AI in the research process. This includes specifying which parts of the work were generated or assisted by AI.

Each piece of information coming from AI must be backed by a specific source, allowing you and others to verify the claims and explore the original research if needed. If there is no backup, ask and verify.

   Researchers must:

  •    Clearly document AI’s role in research
  •    Verify AI-generated outputs with citations
  •    Use AI responsibly while maintaining research integrity

Six C's for effective log keeping

Clarity - Ensure that each log entry clearly defines the purpose of the prompt, the context in which it is used, and the specific goals you aim to achieve. A well-defined prompt is essential for accurate AI output.

Consistency - Keep track of each version of the prompt and any adjustments made over time. This ensures that prompts are used consistently, and you can easily track how each version affects the results.

Comprehensiveness - Document all relevant details about the prompt, including input data, expected output, constraints, and additional parameters, so that others (or you in the future) can fully understand how the prompt was designed and executed.

Change Tracking - Record any modifications made to the prompt and the reasons behind them. This ensures that the evolution of the prompt is well-documented and that you can see the impact of changes over time.

Clarity of Outcome - Keep a clear record of the results generated by the AI, including any issues or unexpected outcomes. This ensures that the output can be evaluated and refined, as necessary.

Collaboration and Continuity - Maintain a log that facilitates collaboration by ensuring others can easily follow the prompt’s evolution and reasoning. This encourages a transparent research process and ensures that prompt management is easily passed on to future team members or collaborators.

Why Data Documentation Matters?

Data documentation is essential for making research transparent, understandable, and trustworthy.

Facilitates Understanding

Accountability - Purpose and Scope

What: Clearly state the purpose of using AI in your research. This helps others understand the goals and intentions behind your work.

Why: Explain why AI is suitable for your research goals. This justifies your approach and builds trust in your methods.

Enables Reproducibility

Traceability - Data Sources

Description: Describe the datasets used, including where they come from and any steps taken to clean or preprocess the data. This allows others to follow your process and repeat your results.

Ethics: Ensure data privacy and ethical considerations are addressed. This protects individuals’ rights and maintains research integrity.

Improves Data Usability

Innovation - AI Models and Techniques

Model Type: Specify the type of AI model used (e.g., neural network, decision tree). This helps others understand the technical approach.

Training: Outline how the model was trained, including algorithms and parameters. This information is crucial for others who want to build on your work.

Communication - Implementation Details

Tools: List the software and tools used. This makes it easier for others to replicate or extend your research.

Supports Improvement

Limitations and Future Work

Challenges: Discuss any limitations encountered during the research. Being open about challenges helps others learn from your experience.

Next Steps: Suggest areas for future research or improvements. This encourages ongoing innovation and collaboration.

Supports Compliance and Ethical Standards

References - Citations

Provide references for all sources, datasets, and tools used. Proper citation ensures credit is given and supports ethical research practices.

Healthy AI Data

In the AI ecosystem, effective data management is crucial for achieving best outcomes. The better we manage and maintain our data, the better the results we can expect from our AI systems. Just like a healthy body requires regular care, attention, and the right habits, so does healthy data.

Data Preventative Care: Preventive measures are essential to maintaining data health. Regular backups and audits help safeguard data against loss and corruption. By implementing these practices, researchers can ensure that their data remains reliable and available for AI applications.

Data Emergency Care: Despite the best preventive measures, data issues can still arise. Data emergency care involves addressing and fixing broken, corrupted, or missing data promptly. Quick and effective responses to data emergencies are vital to minimise disruptions and maintain the integrity of AI systems.

Data Health Check-ups: Regular data assessments, cleaning, and optimisation are necessary to keep data in top condition. These health check-ups help identify and rectify issues before they become significant problems, ensuring that data remains accurate, relevant, and efficient for AI use.

By taking care of your data through these practices, you can ensure that it performs at its best, supporting the reliability of your research. Remember, a healthy AI ecosystem starts with well-managed data.