En son güncellendiği tarih: 20 Şub 2020
When Bloom published the two-sigma problem in 1984, he shared two fundamental findings. The first one is that “everyone can achieve” and the second one is “a student taught 1-1 using mastery learning methods performed above 98% (2 standard deviations) better than a student taught in a conventional class setting.
These results are not very surprising since the lower teacher to student ratio always leads to better results in learning. On the other hand, providing a private tutor for every student is too expensive and well there is not enough teachers to do that ..
So, the question becomes;
Is there a way to achieve the 1-1 teaching?
Educators all over the world research to find an answer to this 30-year-old problem. Everyone is thinking about if the use of technology in the classroom can solve this problem. Despite all the efforts and progress, technology still has not reach to the level of 1-1 tutoring.
What about Artificial Intelligence?
Can AI respond the unique needs of each individual by using learning theories, predictive analysis, cognitive science and machine learning?
Adaptive Learning is defined as “The field—which uses artificial intelligence to actively tailor content to each individual’s needs—draws upon knowledge domains as diverse as machine learning, cognitive science, predictive analytics, and educational theory—to make this learner-centered vision of education a reality.”
Adaptive courses are personalized to each learner.
Individuals’ learning paths are determined by learner inputs, such as performance, prior knowledge, and engagement, that drive an adaptive algorithm.
Bite-size modules with granular learning objectives are very important for adaptive learning. Granular LOs allow the algorithm to pin-point specific concepts a learner may be struggling with, and then provide immediate remediation to target their specific knowledge gaps.
Adaptive Learning uses the mastery-based” learning idea.
Learners must achieve proficiency in order to progress and complete the course; learners can spend however long they need to master concepts.
Adapting Learning also requires adapting testing. Different types of assessments engage different types of learners. Providing assessments in a variety of formats leverages the available diversity in order to better assess a learner’s mastery of the content. Adaptive testing is figuring out each learner’s proficiency or skill-level in as few questions as possible. Item Response Theory” and “Knowledge Space Theory” are used in both adaptive learning and adaptive testing.
HOW IT WORKS?
Adaptive learning provides “Individualized mastery-based” learning by using four theories.
>> Metacognitive theory (The learners learn best when they know what they don’t know.)
>> . Theory of Game Design ( where the idea of levels of the games must be engaging as well as challenging enough so that you continue to work on it but it has to be still archivable so that you cannot loose all the time which means you lose your motivation to continue)
>> . Ebbinghaus Forgetting Curve (AI simply decides the number of repetitions you need to put a skill or a knowledge to your long-term memory and it provides that repetition just before you forget something)
>> Theory of Deliberate (Deep) Practice. Practice and practice until it become a part of the automaticity. Remember, Erickson claims that it takes 10 000 hours of practice to become an expert on something so it rephrases the belief of “experts are always made, not born”.
Intentional focus (practice has to have a specific target)
Challenge exceeds skills
Repetition to Automaticity
Since modules of the adaptive learning are bite-size, It gives the insight down to the granular level of the each learning objective and how the learner interacts with it. AI can use this data to find out what’s working, what's not, what the patterns are across the group and most importantly, use to optimize each learners path to mastery. Since time is the most valuable asset of us, adaptive learning achieves efficiency in time by showing to each learner only what they need to see at only when they need to see it.
Mc Graw Hill is one of the leading publishers becomes a pioneer in adaptive learning with ALEKS. Assessment and LEarning in Knowledge Spaces is a Web-based, artificially intelligent assessment and learning system. ALEKS uses adaptive questioning to quickly and accurately determine exactly what a student knows and doesn't know in a course. You can try ALEKS by checking the Mc Graw Hills website
Overall, I really liked the idea of AI provides individualized mastery-based learning, I have tried to sample adaptive learning module as well, I particularly liked the way it asks the confidence level of the learner at each question. Being able to collect data about accuracy, time spend, and confidence of the learners can give you a great insight to revise your instruction as well.
On the other hand, creating bite-size modules by using granular LO's requires a very detailed work and very long time, and I wonder what happens if the task or the problem involves or requires using more than one objective or higher-level objectives.
I wonder how you think?