Realizing the Full Benefit of Artificial Intelligence for Education

Roy Amara of the Institute for the Future is quoted as saying: “We overestimate technology short-term, and underestimate it long-term.”  How does this apply to AI?  First, we now know that fears of rapid impact on employment were vastly overblown, as they were based on extrapolations of capabilities continuing in an exponential fashion, rather than plateauing.  Furthermore, the progress has been for narrow applications which were, in engineering-speak, “bounded problems” like protein folding most famously.

During this session, we will examine the true abilities and gaps of AI, as delineated in Charles’ co-authored book on Artificial Intelligence in Education, and discuss whether and when these might be resolved in ethical ways.  Armed with that view, we will then address how advances in AI impact education at three levels:

• Curriculum – What we teach for employability
• Instruction – Intelligent Tutoring Systems may be overblown, but narrow slices are possible (e.g., foreign language acquisition)
• Assessments – analytics are a clear fit, but it is mostly data science techniques; we need to define what AI is in this context, above and beyond algorithmic techniques and basic adaptiveness.

Another impact of AI on education is changing what capabilities educators should cultivate and what students need to learn for their future occupation. Over the next two decades, our partnership with computers in accomplishing work roles will typically involve intelligence augmentation—an interweaving of human judgment with machine reckoning— even though a few human jobs will instead be done completely by AI. For instance, AI designed to support the work of teachers might use metrics like how much time students spent on an activity, the number of questions they answered correctly, and the number of attempts in order to evaluate whether particular students need additional instruction in a topic; AI would provide a recommendation to the teacher. The educator then assesses the validity of this suggestion using a host of data points (e.g., students’ levels of engagement overall, personality, performance in adjacent subjects, well-being) and decides how to best craft additional, engaging instruction for the student. Much as how the word processor enhanced human efficiency, productivity, and capabilities, AI can augment human abilities through reckoning—with potentially a great benefit in which the human-AI partnership is capable of more than either person or machine in isolation.