Communications of the ACM
AI or Intelligence Augmentation for Education?
On December 7, 1968, Douglas Engelbart presented a legendary demonstration that showed how newly emerging computing technologies could help people work together. More generally, Engelbart devoted his professional life to articulating his view of the role of computing in addressing societal problems. He emphasized the potential for technology to augment human intelligence. Since that time, many others have developed the concept of intelligence augmentation (IA).
For example, the field of healthcare sees IA as a more ethical framing. One report defines IA as "...an alternative conceptualization that focuses on AI's assistive role, emphasizing a design approach and implementation that enhances human intelligence rather than replaces it." This report argues "health care AI should be understood as a tool to augment professional clinical judgment."
In education, applications of Artificial Intelligence are now rapidly expanding. Not only are innovators developing intelligent tutoring systems that support learning how to solve tough Algebra problems. AI applications also include automatically grading essays or homework, as well as early warning systems that alert administrators to potential drop-outs. We also see AI products for online science labs that give teachers and students feedback. Other products listen to classroom discussions and highlight features of classroom talk that a teacher might seek to improve or observe the quality of teaching in videos of preschool children. A recent expert report about AI and education all uncovered visions for AI that would support teachers to orchestrate classroom activities, extend the range of student learning outcomes that can be measured, support learners with disabilities, and more.
In colloquial use, the term AI calls forth images of quasi-human agents that act independently, often replacing the work of humans, who become less important. AI is usually faster and based on more data, but is it smarter? In addition, there are difficult problems of privacy and security—society has an obligation to protect children's data. And there are even more difficult issues of bias, fairness, transparency, and accountability. Here's our worry: a focus on AI provides the illusion that we could obtain the good (super-human alternative intelligences) if only we find ways to tackle the bad (ethics and equity). We believe this is a mirage. People will always be intrinsic to learning, no matter how fast, smart, and data-savvy technological agents become. People are why agents exist. We think it's important to always have the human in the loop to understand if things are working and if not, to understand why and make creative plans for change.
Today, students and teachers are overwhelmed by the challenges of teaching and learning in a pandemic. The problems we face in education are whole child problems. Why are parents clamoring to send children back to school? It's not just so they can get some work done! Learning is fundamentally social and cultural; enabling the next generation to construct knowledge, skills, and practices they'll need to thrive is work that requires people working together in a learning community. Schools also provide needed social and emotional support. We are simultaneously at a critical juncture where the need to address ethics and equity are profound. In addition to trust and safety considerations, prioritizing the impact, and understanding how it changes interactions and what those implications are for students and teachers is essential when evaluating AI or any technology.
Thus, we recommend a focus on IA in education that would put educators' professional judgement and learners' voice at the center of innovative designs and features. An IA system might save an educator administrative time (for example, in grading papers) and support their attention to their students' struggles and needs. An IA system might help educators notice when a student is participating less and suggest strategies for engagement, perhaps even based on what worked to engage the student in a related classroom situation. In this Zoom era, we've also seen promising speech recognition technologies that can detect inequities in which students have voice in classroom discussions over large samples of online verbal discourse. In some forward-looking school districts, teachers have instructional coaches. In those situations, the coach and teacher could utilize an IA tool to examine patterns of speaking in their teaching and make plans to address inequities. Further, the IA tool might allow the coach and teacher to specify smart alerts to the teacher—for example, for expected patterns in future classroom discussions that would signal a good time to try a new and different instructional move. Later, the IA tool might make a "highlights reel" that the coach and teacher could review to decide whether to stay with that new instructional move, or to try another.
The important difference between AI and IA may be when an educator's professional judgement and student voice are in the loop. The AI perspective typically offers opportunities for human judgement before technologies are adopted or when they are evaluated; the IA perspective places human judgement at the forefront throughout teaching and learning and should change the way technologies are designed. We worry that the AI perspective may encourage innovators to see ethics and equity as a barrier they have to jump over once, and then their product is able to make decisions for students autonomously. Alas, when things go wrong, educators may respond with backlash that takes out both the bad and the good. We see the IA perspective as acknowledging ethics and equity issues in teaching and learning as ongoing and challenging.
By beginning with the presumption that human judgement will always need to be in the loop, we hope that IA for education will focus attention on how human and computational intelligence could come together for the benefit of learners. With IA, restraint is built into the design and technology isn't given power to fully make decisions without a diverse pool of humans participating. We hope IA for education will ground ethics and equity not in a high stakes disclosure/consent/adoption decision but rather in cycles of continuous improvement where the new powers of computational intelligence are balanced by the wisdom of educators and students.
Jeremy Roschelle is Executive Director of Learning Sciences Research at Digital Promise and a Fellow of the International Society of the Learning Sciences. Pati Ruiz ([email protected]) is a Computer Science Education Researcher at the non-profit Digital Promise. Judi Fusco ([email protected]) is a Senior Researcher focusing on STEM teaching and learning at Digital Promise.
No entries found