Opportunity
What the Experts Say
To provide a compelling investment prediction for educational AI, we must look at what experts in the field have published, seeking trends, points of agreement and points of contention.
But before we hear form the experts, we thought it would also be valuable to hear from a high school teacher. This "front line" viewpoint is valuable, particularly for anyone intending to invest in AI that targets the K-12 market.
Click to expand the grey box and learn about this lay educator's perspective...
But before we hear form the experts, we thought it would also be valuable to hear from a high school teacher. This "front line" viewpoint is valuable, particularly for anyone intending to invest in AI that targets the K-12 market.
Click to expand the grey box and learn about this lay educator's perspective...
How Can AI Help You the Most? - A Teacher's Perspective
We asked a Canadian high school teacher what he thinks AI has to offer to schools. Here is what he said:
"As a McKinsey & Company (2019) report said, teachers are not being replaced anytime soon. Where AI can be more than useful - a game changer - is in the "administrivia" of the education industry (augmenting the roles of teacher and administrator):
"As a McKinsey & Company (2019) report said, teachers are not being replaced anytime soon. Where AI can be more than useful - a game changer - is in the "administrivia" of the education industry (augmenting the roles of teacher and administrator):
- It would improve student outcomes through analysis of behaviour trends (missing assignments, absences, etc.) and automatically call in support resource personnel.
- This has massive ramifications as it could catch the large number of students who fall through the cracks in the education system. If AI could analyze absences and missing assignments and then notify parents, counselors, administrators appropriately, it would afford meaningful interventions without the teacher tracking those same data points at the expense of the rest of their roster.
- It would lessen the cognitive load on the teacher, automating menial tasks such as taking attendance, tracking absences and missing assignments, etc. meaning improved working conditions for teachers and better learning environments for all students.
- Compared to many other educational technologies (such as VR headsets, iPads for every student, etc.) this sort of technology would be relatively cheap. No front-facing technologies would be required, other than the method of data entry. Rather, the infrastructure exists on the back-end, or centralized.
- Privacy, security, and appropriate use/management of data would (idealistically) be governed by specific directives in the contract as set out by the school/district/country, as well as federal regulations if this were to be rolled out on a large scale."
When it comes to educational AI, we consider "experts" to be people who have done original scientific research on the topic, have thoroughly studied the topic academically and/or who have been involved in the development of successful educational AI systems.
Our team reviewed a collection of publications (see the references page) from the academic, NGO and business spheres. The expandable boxes below present our summary of expert opinion as to the most prominent types of AI applications for education. The populations directly impacted/targeted by each type, as well as real-world examples of each type, are also provided.
Please look through these different types of AI, and be sure to explore a few of the examples via the given links.
Our team reviewed a collection of publications (see the references page) from the academic, NGO and business spheres. The expandable boxes below present our summary of expert opinion as to the most prominent types of AI applications for education. The populations directly impacted/targeted by each type, as well as real-world examples of each type, are also provided.
Please look through these different types of AI, and be sure to explore a few of the examples via the given links.
Expert-Identified Prominent Applications of AI to Education
Adaptive learning / Intelligent Tutoring Systems
This application is:
This Application Benefits:
Students * Educators
Examples
- A computer-based educational system that changes the type of content, the rate of progression through increasingly sophisticated types of content and other content elements in response to student performance on the system.
This Application Benefits:
Students * Educators
Examples
- Khan Academy
- Rosetta Stone (languages)
- Duolingo (languages)
- Enlearn
- MATHia (Carnegie Learning)
- Querium
- ALEKS (McGraw-Hill Education)
- Dreambox Learning (Math)
- Cognii
- ASU BioSpine Initiative
- OSU Adaptive Learning Program (this also)
- CogBooks Platform
chatbot services
This application is:
This Application Benefits:
Students * Educators * Administrators
Examples
- A computer system designed to simulate conversation with human users, usuly through typed interaction.
This Application Benefits:
Students * Educators * Administrators
Examples
automated essay scoring
This application is:
This Application Benefits:
Students * Educators
Examples
- A computer system that uses a type of AI called natural language processing (NLP) to assess the quality of written content (primarily limited to writing style and language conventions).
This Application Benefits:
Students * Educators
Examples
- Intellimetric
- E-Rater
- Criterion (writing evaluation)
- Revision Assistant (writing feedback)
- Write To Learn (writing and comprehension feedback)
- Grammarly (grammar feedback)
- Chegg Writing (writing feedback)
- Features of MOOC platforms like edX, Udacity and Coursera
automation (administrative tasks)
This application is:
This Application Benefits:
Students * Educators
Examples
- A computer system that automatically performs specific mundane actions, typically to save time for a person who would otherwise have to perform the task.
This Application Benefits:
Students * Educators
Examples
- YouTube (automatic video transcription)
- Diagnostic, instructor-facing features of adaptive learning systems / intelligent tutoring systems
Tracking and Alerting
This application is:
This Application Benefits:
Students * Educators
Examples
- A predictive analytics computer system that uses student behaviour, demographic and academic performance data to predict things like course pass rates, dropout rates and future performance levels.
This Application Benefits:
Students * Educators
Examples
- Early Warning Systems were used in over 50% of U.S. high schools in 2016 (Murphy, 2019)
- Early Warning Systems were used in 90% of U.S. universities in 2014 (Murphy, 2019)
- See also: Regional Education Laboratory Program’s Early Warning Systems Resource Page
Our Analysis
Adaptive Learning is the most prominent type of AI application to education in our research (see the expandable boxes above). It is not surprising that examples are so plentiful, since intelligent tutoring systems have been in development since the early 1980s (Murphy, 2019).
Such a long history means this type of educational AI is more mature than recently developed types, and is ready to evolve into a more sophisticated form. Product area maturity often translates into greater diversity of product offerings and greater depth of product features. These in turn mean greater overall usefulness, wider uptake and superior returns on capital investment.
Furthermore, consider the following expert opinions that surfaced in our research:
These facts suggest that there won’t be any single, “silver bullet” AI application that will transform education anytime soon. We are seeking the best, near future investment and innovation opportunity, so we needed to consider more than simply the most popular or prominent AI applications.
You may recall from the Key Definitions & Concepts area that earlier and simpler forms of adaptive learning use rules-based AI, while more complex forms of adaptive learning use machine learning AI. When determining what path (sequence) of activities a learner should take through an online course, these two forms of adaptive learning work differently:
Such a long history means this type of educational AI is more mature than recently developed types, and is ready to evolve into a more sophisticated form. Product area maturity often translates into greater diversity of product offerings and greater depth of product features. These in turn mean greater overall usefulness, wider uptake and superior returns on capital investment.
Furthermore, consider the following expert opinions that surfaced in our research:
- All existing AI is “weak” or “narrow” AI (Murphy, 2019)
- “Strong” AI (Artificial General Intelligence, or “AGI”) is theoretical and not likely to exist for many decades (Baum et al., 2011; Grace et al., 2018; Vincent, 2018)
- AI applications cannot replace teachers or teacher abilities, but can be very useful in complementing and supporting teachers (Hao, 2019; McKinsey, 2019; Murphy, 2019)
These facts suggest that there won’t be any single, “silver bullet” AI application that will transform education anytime soon. We are seeking the best, near future investment and innovation opportunity, so we needed to consider more than simply the most popular or prominent AI applications.
You may recall from the Key Definitions & Concepts area that earlier and simpler forms of adaptive learning use rules-based AI, while more complex forms of adaptive learning use machine learning AI. When determining what path (sequence) of activities a learner should take through an online course, these two forms of adaptive learning work differently:
Machine-Based Adaptive Learning |
|
Rules-Based Adaptive Learning |
|
Our team believes that machine learning-based adaptive learning systems represent a key part of the most effective and impactful type of AI for education in the near future. With this in mind, one particular real world AI example stood out in our analysis: the BioSpine Initiative at Arizona State University (ASU).
This initiative combines online, machine-learning-based adaptive learning activities with in-class discussions, problem solving, case studies and role-play activities. The approach produced a statistically significant increase in student academic performance, from an average of 76% to and average of 90% :
Student engagement and confidence was also positively impacted. In the first pilot with a Biology 100-level course, pass rates increased by 24% and dropout rates decreased by 90% (Leander, 2019).
This video introduces the CogBoks Platform, the sophisticated adaptive learning system that was used in the ASU case presented above:
This video introduces the CogBoks Platform, the sophisticated adaptive learning system that was used in the ASU case presented above:
Source: CogBooks.com
Our Prediction: Adaptive Redesign 2.0
Rather than looking for a revolutionary, sudden and disruptive AI-based application with dreams of huge early adopter ROI, we recommend investors and education innovators look to mature applications designed specifically for use in conjunction with proven, in-person instructional methods.
After examining expert opinions on the most impactful AI applications to education, our team predicts that one of the most fertile areas for investment and innovation is the emerging approach of combining advanced online adaptive learning systems with in-person active learning classes. This represents an evolution of the “flipped learning” model, and has been referred to as the adaptive redesign of existing programs of study (Educause, 2020).
Some researchers have used Bloom's Taxonomy to help understand the efficacy of the flipped / adaptive redesign approach:
As mentioned above, adaptive redesign is already happening in some universities, and we feel that the very best investment opportunities will lie with a new kind of adaptive redesign that we might call "Adaptive Redesign 2.0".
Adaptive Redesign 2.0 will be characterized by offerings that integrate these elements:
Such convergence of AI capabilities and face-to-face active learning -- if effectively co-designed-- promises to synthesize the very best features of multiple “narrow” AI capabilities into more advanced products that can truly improve student engagement, course pass rates and student performance outcomes. At the same time, it will be leveraging proven teaching methods and evolving traditional teacher-student roles.
Adaptive Redesign 2.0 will not be easy to achieve. The software component is one critical element, but it will fail to gain traction if its interdependence with the course redesign and the active learning components is not properly understood and addressed.
Investors and innovators should look carefully for ventures that not only possess top AI expertise, but also seasoned implementation teams capable of quickly establishing trust with faculty and school administrative groups. Additionally, we strongly recommend that investors and innovators insist on seeing a company's plan to address these common challenges that come with AI-based systems that will involve student and educator data:
Adaptive Redesign 2.0 will be characterized by offerings that integrate these elements:
- Online activities that leverage:
- The most advanced adaptive learning systems
- AI-based natural language processing (NLP)
- AI-based voice recognition
- AI-based vision processing
- In-person active learning activities
- Carefully designed to compliment, reinforce and extend the online experience
Such convergence of AI capabilities and face-to-face active learning -- if effectively co-designed-- promises to synthesize the very best features of multiple “narrow” AI capabilities into more advanced products that can truly improve student engagement, course pass rates and student performance outcomes. At the same time, it will be leveraging proven teaching methods and evolving traditional teacher-student roles.
Adaptive Redesign 2.0 will not be easy to achieve. The software component is one critical element, but it will fail to gain traction if its interdependence with the course redesign and the active learning components is not properly understood and addressed.
Investors and innovators should look carefully for ventures that not only possess top AI expertise, but also seasoned implementation teams capable of quickly establishing trust with faculty and school administrative groups. Additionally, we strongly recommend that investors and innovators insist on seeing a company's plan to address these common challenges that come with AI-based systems that will involve student and educator data:
- Large, high-quality data sets needed to train the systems are difficult to obtain (e.g. training sets should represent the diversity of the application’s target population)
- Statistical models will encode any biases that might be embedded in machine learning training data
- Lack of transparency about a company's AI models (to protect intellectual property) can easily damage public trust in a product
Closing Survey
Please complete this survey as a reflection of your learning through your experience exploring our OER!
How to Access the Survey
> On Laptop or Tablet: Open the expandable grey box below!
> On a smartphone: click here to open in a separate tab!
Please complete this survey as a reflection of your learning through your experience exploring our OER!
How to Access the Survey
> On Laptop or Tablet: Open the expandable grey box below!
> On a smartphone: click here to open in a separate tab!
closing survey (laptop or tablet only)