Greatest hurdles in AI proliferation in Education

Abstract: We experience widespread AI in almost every interaction with information and knowledge today. But there is one information-heavy industry that is consciously struggling in adopting a bare minimum amount of it: Education.

It's easy to dismiss existing educational institutions for their inability to bring in the wonders of deep learning and knowledge retrieval. But if we dig a little deeper, we discover that there are 5 key factors that prevent teachers and students from capturing the value generated by scalable predictions beyond randomized control trials. These factors are:

- Reconciliation of competency representations
- Codification of curriculum and pedagogy
- Economic incentives to deeper AI research
- Holistic capturing of student performance
- Sample efficient machine learning

This talk dives into each of these 5, and offers suggested pathways on collective actions we need to take to solve them in the next decade.

Bio: Varun Arora is a senior AI product engineer and manager at Baidu USA, where he works with AI researchers on productization of deep learning technologies. He is also involved in the development and adoption of Baidu’s deep learning platform, PaddlePaddle. Previously, Varun was the CEO of YC-backed education technology company OpenCurriculum, where he worked with thousands of teachers and hundreds of education administrators in improving curriculum in K–12 classrooms. He has also worked for Inkling, the UN, and a One Laptop Per Child deployment, and advises several companies in education technology. He holds a bachelor’s and master’s degree from CMU.