Inside the Black Box of Adaptive Learning Technology
BAD PRESS
In my previous post, I mentioned that the media
has had much to do with our culture's fear of technology--technophobia, and fear
of being spied—scopophobia. The news media has had an impact creating fear of
technology. According to writer Dwayne Day, “’2001: A Space Odyssey’ movie has
been the most culturally and socially influential science fiction film ever
made.”
Why? In the movie, astronauts are on a mission to Jupiter and the spaceship’s computer, HAL9000, kills all but one crewmember, who then unplugs HAL. Perhaps you’ve heard of the 1938 debacle where Orson Welles dramatized a radio program that was produced to sound like real news of a Martian invasion in New Jersey. This radio broadcast was intentionally formatted and dramatized to cause nationwide hysteria (Schwartz).
Why? In the movie, astronauts are on a mission to Jupiter and the spaceship’s computer, HAL9000, kills all but one crewmember, who then unplugs HAL. Perhaps you’ve heard of the 1938 debacle where Orson Welles dramatized a radio program that was produced to sound like real news of a Martian invasion in New Jersey. This radio broadcast was intentionally formatted and dramatized to cause nationwide hysteria (Schwartz).
What about George Orwell’s book, 1984, where the plot contains official deception, secret surveillance, brazenly misleading terminology, and manipulation of recorded history? Fast-forward to May, 2013, when Edward Snowden leaks to a Washington Post reporter that the NSA has a surveillance program called PRISM, which collects data from Facebook, Microsoft, Google, and other social media sites. Author Ian Crouch notes the spike in book sales for 1984 since Snowden fled to Russia and Hong Kong and wrote, “The technological possibilities of surveillance and data collection and storage surely surpass what Orwell imagined.”
TRACKING AND ARTIFICIAL INTELLIGENCE
Adaptive learning technologies—ALEX, Knewton and MyLab, by McGraw Hill and Pearson—made their classroom debut between 2011 to 2013. Parents, students, and teachers were skeptical of adaptive learning, perhaps due to the media’s negative influence on technology and fear of change. Seven years later, adaptive learning technology continues exist as a 21st century classroom tool. The factor missing from adaptive learning technology’s wide-spread acceptance: buy-in. According to Oremus (2015), adaptive learning technology is only in its Model T phase. Adaptive learning technology is so new, there isn’t enough data for school districts to justify the cost of switching over to adaptive learning technologies in the classroom. Others claim their dislike of the data collected by adaptive learning technology. Regardless of whether we dislike or agree with it, the technology we use every day collects information about us by tracking our Internet usage, our queries, the GPS on our cell phones, Google Maps, and Waze; credit card purchases (what we buy and where we buy it), traffic cameras, voice recognition systems (Siri and Cortana) on cell phones and personal assistants like Google Home and Amazon Alexa, and facial recognition systems at airports, sporting arenas, ATMs, and on social media sites such as Facebook. On a commercial airline flight, there is only seven minutes of human steered flight—at take-off and landing. The rest of the flight uses artificial intelligence—AutoPilot. It seems logical to conclude that if we are using technology, then we are inevitably tracked, traced, and databased. While the social and ethical issues regarding the topic of Big Data are important, they are not the focus of this post. In this post, the past has been examined to evaluate possible reasons why our culture feels the way it does with regard to data collection and artificial intelligence. Next, we’ll see how digital adaptive learning technology collects data to help students learn in the most efficient and effective manner possible.
ADAPTIVE LEARNING DEFINED
Edsurge defines digital adaptive learning as the educational tools that can respond to a student’s interactions in real-time by automatically providing the student with individual support. Common misconceptions of what adaptive learning is include (1) an instructional strategy, (2) the use of software, (3) differentiated instruction, (4) personalized instruction, and (5) individualized instruction. Digital adaptive learning deals with real-time data collection, automatic responses, and response or redirection for students based on their unique choices.
Note that digital tools that don’t collect data in real time are not adaptive. Tools that don’t provide support in real-time are not adaptive. Tools that mark answers correct or incorrect, then provide a single path for learning regardless of the response are not adaptive.
STUDENTS TAKE RESPONSIBILITY FOR THEIR LEARNING
Many teachers don’t want to change their ways, and many want to see evidence that digital adaptive learning technology really works. There are privacy and budgetary concerns, and questions about the viability of companies that produce and support the digital adaptive learning tools. On the other side of the scale is delivering engaging content that makes students more enthusiastic learners. Until teachers change the way they work and more evidence is collected, 21st century adaptive learning will touch only 20% of the nation’s K12 students.
DIGITAL ADAPTIVE LEARNING TECHNOLOGY
Adaptive learning happens in three places: in content, in sequencing, and in assessments. Look at the chart to see how adaptive learning technology respond differently to two students.
Adaptive learning technology collects data, analyzes the data, then adjusts the content the student will see next so students learn as efficiently and effectively as possible.
ARTIFICIAL INTELLIGENCE: WHAT IS IT AND HOW DOES IT WORK?
Artificial intelligence—called AI is defined as, “computer systems designed to interact with the world through capabilities (visual perception and voice recognition), and intelligent behaviors (taking the most sensible action to achieve a goal after assessing the available information) that we would think of as essentially human.” (Luckin, Holmes, Griffiths, & Forcier, 2016). People worry that AI will reach the point of singularity—the point where an AI-powered computer or robot could improve or redesign itself, and leading scientists, such as Stephen Hawking have expressed concerns about “the potential downsides of AI becoming too clever” (Luckin, et. al., 2016). While AI is becoming more sophisticated, general AI does not exist; only domain-specific or narrow AI exists.
For more than 30 years, the application of artificial intelligence to education (AIEd) has been the subject of academic research. AIEd is what gives adaptive learning technology its ability to adapt teaching and learning approaches and materials to the capabilities and needs of the individual learners. AIEd, sometimes called the ‘black box of learning’ is what provides the understandings of how learning actually occurred. The two parts of AIEd are (1) knowledge about the environment it is processing and (2) algorithms to intelligently process the knowledge.
The AIEd needs to know about the knowledge and expertise of teaching, or the pedagogical model (allowing a student to make a mistake before showing the right answer, feedback triggered by the student’s responses, and assessment to measure learning). The second model the AIEd needs to know is the domain model, which includes the rules and procedures for the specific subject (math, history, etc.). Finally is the learner model, which is data on the student’s previous attempts at this and prior work, and the student’s engagement in learning (time on task and other measurements). The learner’s responses are constantly fed back into the learner model, which makes the model richer. This is how AIEd ensures that the student’s learning experience effectively supports their learning. In order to do this, AIEd gathers large amounts of data to dynamically improve the pedagogy and domain models to help teachers refine the processes of teaching and learning.
From the explanation and model pictured, the data collection is essential input into all three models in the AIEd of adaptive learning technology. The AIEd, using the collected data, is how the “adaptive” happens! By collecting this data, even the data from the tracking discussed earlier in this post, we are creating an ocean of digital data. It is estimated that in the current year (2018) there will be 7.6 billion people with 25 billion internet connected devices (Luckin, et. al., 2016). Information from devices is communicated to applications discussed earlier in this post, and mothers. Adaptive learning technology collects data in a similar fashion, with respect to the individual items on an activity and time; more than traditional data collection of a duration of an activity and an activity score.
The last thing to mention is the teachers. AIEd can perform teacher activities such as grading and record keeping, but AIEd can not replace teachers. Instead, with adaptive learning technology, teachers are freed to dedicate their energy to be creative and take learning up a notch (Luckin, et. al., 2016). With adaptive learning technology and AIEd, there is no doubt the role of the teacher in the classroom will change.
ADAPTIVE LEARNING DEFINED
Edsurge defines digital adaptive learning as the educational tools that can respond to a student’s interactions in real-time by automatically providing the student with individual support. Common misconceptions of what adaptive learning is include (1) an instructional strategy, (2) the use of software, (3) differentiated instruction, (4) personalized instruction, and (5) individualized instruction. Digital adaptive learning deals with real-time data collection, automatic responses, and response or redirection for students based on their unique choices.
Note that digital tools that don’t collect data in real time are not adaptive. Tools that don’t provide support in real-time are not adaptive. Tools that mark answers correct or incorrect, then provide a single path for learning regardless of the response are not adaptive.
STUDENTS TAKE RESPONSIBILITY FOR THEIR LEARNING
Many teachers don’t want to change their ways, and many want to see evidence that digital adaptive learning technology really works. There are privacy and budgetary concerns, and questions about the viability of companies that produce and support the digital adaptive learning tools. On the other side of the scale is delivering engaging content that makes students more enthusiastic learners. Until teachers change the way they work and more evidence is collected, 21st century adaptive learning will touch only 20% of the nation’s K12 students.
DIGITAL ADAPTIVE LEARNING TECHNOLOGY
Adaptive learning happens in three places: in content, in sequencing, and in assessments. Look at the chart to see how adaptive learning technology respond differently to two students.
