This is the second blog on the new TrustEd Apps Rubrics (self-assessment tools) developed by the 1EdTech community.
TrustEd Apps Generative AI Data Rubric: 5 Questions to Ask Before Implementing an AI Tool
In the past year, artificial intelligence has brought out just about every emotion in the world of education. Concern, excitement, inspiration, fear, the list goes on. In response to these various reactions, many organizations and businesses worked diligently to provide some type of guidance around AI, including an executive order from President Biden in October.
The president’s executive order is also a signal of future regulations rather than simply guidance.
As a collaborative, innovative, and forward-thinking organization, 1EdTech members responded in a similar manner, looking for not just best practices but practical tools that can help educators make decisions around AI and help suppliers provide tools that meet educational needs.
It will be a long process to cover every aspect of AI, but our board felt data privacy in generative AI was one of the most pressing issues that needed to be addressed.
With that in mind, a collaborative group of AI experts from our membership worked together to create 1EdTech’s TrustEd Apps Generative AI Data Rubric. The rubric is designed to serve as an informative and educational tool for suppliers and educators to understand some of the most important questions around privacy and what educators should expect from reliable suppliers.
“With the rapid pace of advancement around generative AI, it’s imperative that we take proactive steps to ensure the safety, privacy, and accessibility around these tools,” said Ryan Lufkin, vice president of global strategy at Instructure. “This effort by 1EdTech will ensure a level of transparency that is fundamental to the long-term trust in generative AI within education.”
The rubric is made up of five questions that suppliers will answer for themselves and publish for any potential customer to review and consider.
The questions ensure potential customers understand the following:
- If they will know AI is being used in a tool
- If there are opt-in or opt-out clauses when it comes to AI use
- If third parties are involved
- Information on how the AI model was trained and how it continues to be improved
“There are nuances to how AI will use, store, and present data it has ingested that is not as clear-cut as access rules we have today for databases, file systems, etc.,” said Eric Galis, chief information security officer for Cengage. “Much like when you or I read a book, we don’t store the book verbatim in our brains but remember parts of it. AI’s use of tokens results in a similar pattern. Due to the conversational nature of GenAI interactions, users may let down their guard and share information they wouldn’t elsewhere online. Therefore, we cannot rely solely on current privacy practices to ensure that private data does not make its way into training corpuses.”
Rubric collaborators stress that the rubric is primarily an educational tool to help all stakeholders consider the privacy issues surrounding AI.
“The AI rubric benefits edtech suppliers by providing a standardized evaluation framework, ensuring all suppliers adhere to the same high standards and prioritize responsible and ethical AI in their AI offerings, thereby streamlining assessment, enhancing product quality, and fostering continuous improvement,” said Shivani Stumpf, chief product and innovation officer at PowerSchool. “K-12 school districts and higher education institutions will benefit by providing a consistent and transparent evaluation framework, ensuring that educational technology solutions adhere to high standards, prioritize student data security, and follow responsible AI practices, thereby promoting trustworthy and effective integration and adoption of this technology that has the potential to transform education, enhancing overall learning experiences, and safeguarding the privacy and well-being of students.”
The TrustEd Apps Generative AI Data Rubric will join 1EdTech’s Data Privacy Rubric, the recently released Security Practices Rubric, and the upcoming Accessibility Rubric in 1EdTech’s TrustEd Apps Management Suite. Until then, suppliers are encouraged to review and consider their own policies before self-reporting.
1EdTech will continue to work with members to address and set expectations for AI use and develop a more comprehensive instrument to assess the use of data in AI, including data sources, data validity, bias, and more.
Institutions looking for additional guidance right now can access the AI Institutional Preparedness Checklist to start developing responses to AI and initiate conversations within organizations.
About the Author
Beatriz Arnillas started her career as an art professor. She became interested in technology as an opportunity to deliver blended and online courses for non-art majors. Subsequent positions included higher education administration (assistant chair, director.)
Beatriz switched to adult education when she was Head of Education Production and Outreach at the University of Houston's College of Technology (COT). She developed and managed the COT online and graduate course design team, industry certifications, and continuing education programs. After U. of Houston, Beatriz was a Global Sr. Instructional Designer in the oil and gas industry. She returned to education as Director of IT and EdTech for the Houston Independent School District (HISD). After Houston, Beatriz worked in the private sector as the Sr. Global Education Advisor of Itslearning, Inc., and Sr. Instructional Designer in AWS Training and Certification before joining 1EdTech.
Beatriz has delivered keynotes and consulted with education leaders in the US, Germany, Norway, South Africa, and China. Before joining 1EdTech as Director of Digital Curriculum Innovation in Oct. 2021, Beatriz designed Architecting and Adv. Architecting on AWS courses in the AWS Training and Certification Org (global AWS training.)