Enhanced Learning Through Machine Learning

DreamBox, an ed-tech company based in Seattle, uses machine learning to create an adaptive platform that teaches k-8 math.

Machine Learning, The DreamBox Solution:

DreamBox, an education technology company based in Seattle, “was founded with the mission to transform the way the world learns by using intelligent adaptive technology to dynamically and continuously personalize each student’s learning experience”[1]. In its essence, DreamBox uses machine learning technology to teach math to k-8 students through a personalized online platform. This platform is meant to be supplementary to the classroom experience. Dreambox uses data analysis to support its teaching process through Adaptive Learning Technology. According to the DreamBox website, “the Intelligent Adaptive Learning technology tracks each student’s interaction and evaluates the strategies used to solve problems. It then immediately adjusts the lesson and the level of difficulty, scaffolding, sequencing, number of hints, and pacing as appropriate. This allows students, whether struggling, at grade level, or advanced, to progress at a pace that best benefits them and deepen conceptual understanding.”[2] DreamBox is solving a gap in the education industry by improving the math learning experience for elementary and middle school students. DreamBox CEO, Jessie Woolley-Wilson, says personalized learning can let “students who are behind catch up and…students who are ahead move forward. If the software notices that a student is using an inefficient procedure to solve a particular problem, it might cut in with a targeted lesson.”[3]  Machine learning is particularly important in improving DreamBox’s processes, as it allows to 1) collect large amounts of data on  how they interact with the math platform, and 2) based on this data, create a unique experience for students that could greatly impact their academic path. The more users interact with the platform, the more the platform learns about the users. DreamBox is solving a problem that requires prediction to develop the best possible learning track for a student by identifying patterns in responses and interactions with the platform. It is also serving to prove to educators that technology solutions can enhance the classroom experience.

Next Steps:

DreamBox recently received a $130 USD million investment from private equity fund TPG[4]. These funds will be used in the coming years to increase access to math education around the world. [5] In the short term, DreamBox will be focused on expanding its userbase and testing the platform in other geographies. According to a Seattle times article “The company has had pockets of success…but what [Woolley-Wilson] really wants is significance – or proof and ability to expand the software to any school with repeated positive results.” [6] The short-term focus of DreamBox will be to prove the significance of results to date by expanding its userbase and geographic footprint, specifically into Asia, the Middle East, Africa, and Latin America.[7] In the long term, Woolley-Wilson is looking to change how educators interact and feel about technology. She claims: “the skepticism educators feel about learning technology is justified. We have as an industry overpromised and underdelivered. And what that translates to is when there is something that actually works, that is engaging, that is personalized, and that is efficacious, people don’t believe it.” Over the long-term, she hopes to “change the narrative.”[8]

Recommendations:

DreamBox’s pursuit to prove significance in their results is extremely important for the near future. A Harvard study on the use of DreamBox on student achievement found that while “DreamBox progress measure was positively associated with achievement gains on state tests and interim assessments”, “the evidence of causal impact…is encouraging but mixed”[9]. It is critical for DreamBox to prove the causal relationship of their math platform and performance results. However, I would recommend DreamBox to begin exploring other fields in which to apply their technology. Even if DreamBox proves to be successful in the math space (which is a subject with little subjectivity in the types of responses a student can give to a question), their algorithms might not be applicable to other areas. With growing competition in the ed-tech space, DreamBox should be thinking ahead to become an education solution in more than one field. This would also aid Wolley-Wilson’s objective to change the education narrative more broadly.

Additional Questions:

Upon reading about DreamBox and its CEO’s ambitions, some questions are raised: 1) Can machine learning in education be as successful without the support of a well-trained teacher and, therefore, could machine learning be the solution for regions in the world that lack sufficient teachers? Or will this type of digital solution always require a human component?; 2) Are there fields of education where machine learning would not add significant value or are all fields in education destined to eventually use machine learning in some way?  (758)

 

[1] Clare McGrane (2018) DreamBox Learning raises $130M, adds former U.S. Education Secretary to board, GeekWire. Available at: https://www.geekwire.com/2018/dreambox-learning-raises-130m-adds-former-u-s-education-secretary-board/ (Accessed: 13 November 2018).

[2] What is Intelligent Adaptive Learning? (no date) DreamBox Learning. Available at: https://www.dreambox.com/intelligent-adaptive-learning/ (Accessed: 13 November 2018).

[3] Benjamin Herold (2018) ‘What Does Personalized Learning Mean? Whatever People Want It To – Education Week’. Available at: https://www.edweek.org/ew/articles/2018/11/07/what-does-personalized-learning-mean-whatever-people.html (Accessed: 13 November 2018).

[4],[5],[6] Rachel Lerman (2018) ‘DreamBox Learning gets $130 million for math education software | The Seattle Times’, 31 July. Available at: https://www.seattletimes.com/business/technology/dreambox-learning-gets-130-million-for-math-education-software/ (Accessed: 13 November 2018).

[7],[8]  Ainsley Harris (2018) ‘DreamBox Learning’s adaptive math lessons get a $130 million boost’, 31 August. Available at: https://www.fastcompany.com/90208629/dreambox-learnings-adaptive-math-lessons-get-a-130-million-boost (Accessed: 13 November 2018).

[9] Jon Fullerton (2016) DreamBox Learning Achievement Growth, Key Findings Report: DreamBox Learning Achievement Growth. Available at: https://cepr.harvard.edu/dreambox-learning-achievement-growth (Accessed: 13 November 2018).

Previous:

Waymo: The future of (not) driving

Next:

Unlocking the Power of Open Innovation in Disaster Relief

3 thoughts on “Enhanced Learning Through Machine Learning

  1. Patricia – thank you for the interesting read!

    To your first question; I am of the opinion we will always need teachers in the classroom. It is the responsibility of teachers to provide a well-rounded education to students. While this education should cover teaching traditional hard math and language skills, it also involves inspiring students and helping to diagnose social and emotional causes of learning difficulties. To take an example, for a child whose parents are going through a divorce, a human-teacher can recognise this and provide support and inspiration that a machine can’t. Furthermore, I worry that the lack of human-contact resulting from machine-only teaching could curtail the development of children. Such concerns have also been raised relating to the use of ‘robot nannies’, with early research suggesting it could lead to developmental impairments among younger children [1].

    As a result, I agree that machine learning has application in education, but I believe this is limited to acting as a support-tool for teachers (e.g., for grading, teaching of simple concepts and record keeping). Given what we know about machine potential, McKinsey estimates only 27% of education activities could be automated in the future [2]. While this number may increase as technology develops, for the reasons mentioned above, I don’t see it ever hitting 100%.

    [1] N. Sharkey, A. Sharkey, “The crying shame of robot nannies: an ethical appraisal,” Interaction Studies: Social Behaviour and Communication in Biological and Artificial System, 11(2) (2010): 161-190.

    [2] M. Chui, J. Manyika, M. Miremadi, “Where machines could replace humans—and where they can’t (yet),” July 2016, https://www.mckinsey.com/business-functions/digital-mckinsey/our-insights/where-machines-could-replace-humans-and-where-they-cant-yet, accessed November 2018.

  2. While Dreambox sounds like a fantastic company with potential to add significant value to the classroom, I worry about over-reliance on machines to teach skills that would otherwise be taught in a classroom. In addition to hard skills, children rely on their teachers and peers to learn social skills, group dynamics, and coping mechanisms. Like Colm, I think Dreambox could be a wonderful tool to add to a teacher’s portfolio, but it should not replace a teacher entirely.

  3. Really interesting company! I agree with the points raised by Amina. I think the interaction between students and teachers can inspire an important, early interest in education that cannot be replicated by a machine. However, I think that DreamBox could be an extremely useful supplementary tool in personalizing the education across a set of students that do not have the same capabilities. Essentially, it could ensure that less students fall behind, as they receive the adequate level of help based on their level. I see a huge value of this service, like you mentioned, in regions with fewer quality teachers and could see it being supplemented with AR/VR potentially in the future as well.

Leave a comment