7 Questions on Anthology's Approach to AI
CT: Is there a learning curve for educators and administrators in understanding how best to use AI tools and how to interpret their output?
White: Interacting with and creating prompts for generative AI is a skill, and most people are still at the very early stages of developing that skill. The beauty of certain tools like Anthology's AI Design Assistant is that it helps instructors create more engaging learning experiences more quickly without requiring those skills. The capabilities are seamlessly integrated as part of the existing workflows and are available in the context where they are needed.
For example, an instructor may want to create a grading rubric for an assignment. Instructors can continue to create their own rubrics from scratch, but with the AI Design Assistant they now get an option at the point of creation to have Blackboard Learn suggest a grading rubric based on the context of the course and the assignment. The instructor can then review, edit, and decide whether to accept the suggested rubric.
This pattern of easy-to-use contextual suggestions is applied consistently to all the AI Design Assistant capabilities, which is very intuitive and has a very small learning curve.
CT: How do you see AI tools impacting the student experience?
White: It's important to take a careful and thoughtful approach to implementing AI tools across any aspect of the educational lifecycle, but especially as it relates to the student experience. Our initial focus for AI-related tools is centered around inspiring instructors, empowering advisers, and supporting institutional leadership by providing meaningful efficiencies that allow to them to deliver more engaging student experiences with human oversight.
AI technology is still in its infancy. We cannot forget about AI risk factors, such as the potential for bias and inaccuracies. This is why we developed our Trustworthy AI Framework and believe that it's critically important that instructors and faculty are always part of the review and oversight process.
There has been a lot of focus on the inclusion of AI detection in learning management solutions. It is our opinion that reliable detection is not yet a fully viable approach. AI detection can provide a false sense of security and it can disadvantage students with disabilities and those who are learning outside their native language. Instead, an emphasis should be placed on embracing authentic assessment. With AI able to distill existing information with such efficacy, assessment needs to focus on critical thinking, personal perspectives, and self-reflection rather than solely on the accrual of knowledge.
AI has the potential to further reimagine the student experience with improved personalization and empowerment. Institutions can create or design programs that are more adaptive to skill development and employer needs with greater efficiency being responsive to both the learner and the community.
CT: What does that Trustworthy AI Framework consist of?
White: Anthology has been doing policy work on the ethical use of AI and data since 2018. And, with recent acceleration in generative AI, we released our Trustworthy AI Program and Policy to ensure that approach is instilled in all of our initiatives. The Trustworthy AI Framework has been developed in collaboration with clients, and defines several principles that we adhere to:
- The institution, the instructor, and the instructional designer are always in control. They need to explicitly opt into these capabilities.
- Any suggestions made by the system need to be explicitly reviewed and accepted by the instructor or instructional designer.
- We address copyright and data privacy concerns. No institutional data is used for training models.
- AI comes with risk such as potential for bias and hallucinations. We fervently believe that a human always needs to be in the middle to review and accept. AI tools should not be directly used by students without any human intervention or in high-stakes workflows such as grading.
- We provide an institutional audit trail and reporting to allow the institution to easily see where AI suggestions have been incorporated.