Educational Leaders Without Borders Biennial Meeting University of
Educational Leaders Without Borders Biennial Meeting, University of West Attica, June 2 -4, 2021, 4 -7 pm Eastern Europe Time, Athens, Greece. Panel Discussion: Discourse on Artificial Intelligence and the Role of Ed Leaders Predictive analytics and artificial intelligence in education: helpful or scary? Daniel W. Eadens, Ed. D Danielle M. Eadens, Ph. D Vasily Yurin, MA
Do educational leaders understand the ramifications of Big Data in education? • What is Big Data and what are the Benefits and Risks and benefits? (answer in the chat) • Benefits vs. risks (Janssen et al, 2020) • AI algorithms can lead to unfair results • This is not only unethical, but immoral and illegal • Why better KNOWING issues & benefits of using Big Data is important?
Identifying the identity TED Talk, What will a future without secrets look like, Alessandro Acquisti’s experiments showing how computers can recognize individuals in photos by being sent to the Cloud where nearly a million faces are recognized in seconds. His study found 1 out of 3 results found successful recognitions • You could look at anyone w/ Google Glasses and have 7 data bits about them revealed. This could then determine decisions about what they will be allowed to have, buy, or borrow, frightening! (Acquisti, 2013).
Data and discrimination • In Big Data, the Internet, and the Law, Latanya Sweeney • A cross-country study of 120, 000 Internet Search-Ads found repeated incidence of racial bias. • Google Ad. Words - buys made by companies that provide criminal background checks • Twitter networks found discrimination at Princeton’s tutoring services against poor and minorities • 2009 US Sup Court case Ashcroft Vs. Iqbal • 2015 Texas Department of Housing vs Inclusive communities
Recording and Evaluating data • Ed Leaders are responsible for decision making • Need better understanding- usage of big data • How data is input, standardized, and analyzed • Must have Clear Demarcations/Benchmarks • We need Better Clarity & Equity of Data usage • Collective agreement about algorithms • No secrets within algorithms • Learning analytics helps us underset student learning and needs • Data mining – IS THE new future for Education (Johnson, 2014), • Ethical issues
Protecting data & Approaches to that Protection • Protecting Personally Identifiable Information (PII) is challenging • Using Governance and trusted frameworks w/ authentications can help avoid misuse, hacking, & illegitimate data usage • Privacy Laws are outdated - people need control over the information about them and privacy laws (Sprague, 2014). • The Fourth Amendment (balance) seems to no longer protects the privacy of the citizens, due to new ways of collecting data do not fit the definitions of the Fourth Amendment • Hard to regulate how info is collected so Lawmakers are creating new doctrine, use restrictions, to shift regs on how the data is collected/used • There are significant generational differences in how people approach and feel about data privacy and protection that must be considered
Data bias • Educational leaders and data bias Unintentional biases can be embedded into the algorithms • Individual review boards • Separating protected populations • Fair Test with Python scripting
Algorithm audit • The era of blind faith in Big Data is OVER! • Value-Added Model-secret (VAM) secret algorithm model that does teacher evaluations. • Need for data integrity checks • Use the behind-the-curtain audition methods • Understand that NO algorithm is perfect • Algorithms Can Learn
Algorithm audit • Applicability of Individual profiles • Merging the initial data set with similar profiles • Misrepresentation by individuals themselves • Kenyon College commencement, speaker Nate Silver • Objective evidence data findings • Never stop questioning data
Conclusion and Implications Educational Leaders Should: • Understand how data is collected and used • Establish Boundaries and benchmarks • Create Thresholds and triggering mechanisms • Careful with Tracking academic behavior • Trusted frameworks, authentications, governance • Ongoing integrity checks of the algorithms • Use behind-the-curtain audition methods • Understanding that algorithms are not perfect • Considering errors and feedback loops Then, we are more efficaciously and ethically using Big Data & AI in learning environments to promote student success.
Contact Information Daniel W. Eadens, Ed. D. Associate Professor Educational Leadership Phone: 407 -823 -4751 Daniel. Eadens@ucf. edu Danielle M. Eadens, Ph. D Lead Faculty, Integrative Studies Lecturer, Interdisciplinary Studies danielle. eadens@ucf. edu Vasily Yurin, MA Educational Leadership Higher Education vasily. yurin@ucf. edu
- Slides: 11