What are the potential risks and negative impacts of implementing this AI system?
How will AI systems be used?
What is needed for accuracy?
How has the AI model been trained?
What data sources will be relied upon by an AI model to perform work?
How might it impact data privacy and security, especially with regard to sensitive information?
What data will be provided?
What sensitive data will be provided? (Note: Sensitive data includes but is not limited to: Personally Identifiable Information (PII) such as names, demographic data, geographic data, biological data, etc.; and intellectual property information)
How will the data that is provided be protected?
How will data that is provided be used?
What data will be produced?
How will data that is produced be used?
Are there any uses of our data by provider other than college use? For instance, will data be used for training an AI model in any way?
Vendor must contractually agree to handle all data in accordance with any applicable privacy laws and regulations
What are the implications of this system with respect to fairness, accountability and transparency?
How has vendor addressed bias in the use data or the creation of the output?
Vendor should warrant it has not infringed on third-party intellectual property rights in the AI system inputs or outputs
Does the system appear in the Augustana AI Systems Database with any flags or warnings?
If so, how will you address these warnings while using the system?
What concrete steps will be taken to mitigate identified risks?
Implementation considerations
Have Augustana's AI Professional Development Trainings been reviewed?
Consider transparency and whether the use of this system requires disclosure
What training will be required for safe & effective use? Does this training material already exist or will it need to be created?
How will the accuracy and fairness of outputs be verified?
What human oversight mechanisms need to be established to ensure safe and fair use?
