6 Ways Machine Learning Will Evolve Classrooms in 2021

Written by Nickey Khemchandani

As a technologist, I spend a lot of time with my ear on the ground for the latest happenings around the EdTech Industry.

One of the trends I’ve seen in 2021, is the use of Machine Learning being implemented in EdTech tools and teaching practices.

This is an interesting development because, by definition, machine learning uses artificial intelligence to improve upon itself. When utilized by educators in the classroom, it opens up a number of opportunities for schools to optimize their curriculum and teaching strategy with detailed insights.

In this article, I’ve identified six interesting developments and challenges from my research that may help to guide educators through 2021.

1. Predict when students will struggle with a type of concept

Are you about to start teaching a topic focused on critical thinking?
 
Will a particular student struggle based on previous critical thinking based topics? – These are the type of answers Machine Learning can provide to teachers.
 
Using historical assessment data, many EdTech companies are able to predict when a student may need more help.

2. Which technique works best for a student

The term personalized learning has been around now for over 10 years but we are finally going to see this implemented this year.

To help us get there, Machine Learning is using various information points to identify a student’s learning style, as you can see in the diagram below.

A large task for educators will be to consider tagging individual learning modules. This will allow algorithms to better understand what modules worked better for which student using relationships between the tags.

It should be noted that GDPR’s maturity and accessibility have played a big part in allowing access to relevant student data and will continue to do so.
img
Learning styles being identified by student behavior

3. Teachers building their own learning models

Machine Learning relies on its training data to learn how to navigate data.
 
To date, EdTech companies are training their own Machine Learning systems. This year we will start to see learning models provided by teachers.
 
Instead of using preset and student data, teachers will start providing their own data to Machine Learning. This will allow these tools to become effective teaching assistants in a sense.
img
Teachable Machine is a great project that can be the start of how each of us will train our own Machines

4. Automated testing of curriculum

Curriculum designers are often looking to run test groups to improve their curriculum designs. With the aid of Machine Learning, data can be used to enhance areas such as curriculum flow. Does your lesson have the right amount of reinforcement? The system can even recommend quizzes where student engagement could increase.

A method commonly used by many EdTech organizations is the A/B test, which samples the curriculum with two groups and measures the effectiveness.

With Machine Learning and the added benefit of digital learning, we will also be able to measure student impact on assessment, engagement rates, the effectiveness of reinforcement techniques, and more.

In addition, the speed at which Machine Learning can analyze data is far greater than what we can achieve without its help. This will be a great step in the direction of truly personalized learning.

img
Machine Learning can easily conduct A/B testing used to determine a winning solution

“We help thousands of teachers at BSD Education with our ready-made digital curriculum and projects” – Nickey Khemchandani

5. Body language detection and Audio analysis on video calls for behavioural health

Remote learning is now common practice. Teachers are facing the difficult task of identifying engagement or interest via the student body language on a video call.

It was not uncommon to hear the sound of disengagement or stress in student voices when teaching online. One of the areas Machine Learning was able to step in and help was to highlight “stress” indicators in students’ voices in a lesson.

Years ago, I read an amazing research paper by Ishan Behoora and Conrad Tucker from Carnegie Mellon University [https://bit.ly/3npEB1S] explaining how Machine Learning can classify the emotional state of designers in real-time. This got me keeping tabs on this space for how it can be utilized in Education.

As video calls become a norm in education, expect to see real-time detection of student engagement and attention tracking coming soon.

img

The following is from a sample class we conducted with a technology partner. The voice of a pre-teen [below 13] was analyzed to identify stress levels. (Voices of pre-teens are easier to analyze since often their voice patterns sound similar and there are fewer similarities to an adult voice).

img
Zoom already allows hosts to identify participants that are not actively on the zoom tab or session

6. Assisted grading of non-binary assessments

We have all seen multiple-choice questions being graded using machines. They work great and have been an incredible help for teachers.

With Machine Learning tools used in popular plugins such as Grammarly or the Hemingway Editor, it was only a matter of time before essay writing was also supported.

Research papers are already sharing promising developments and improvements in this space with the inclusion of Machine Learning.

img
The Hemingway Editor is one of my favourite tools when creating a curriculum. http://www.hemingwayapp.com/

“Machine Learning tools need to be accessible by students as well. These tools can help students solve problems while they are encountering them. This is one of the goals technology aims to serve in Education.”
– Nickey Khemchandani

At BSD Education, supporting educators/teachers is at the core of who and what we stand for.

With so much to look forward to as technology develops in education, it will be critical for the education community to support teachers throughout these transitions.

We are excited to hear how more schools incorporate machine learning in 2021.

About Nickey:

He is the co-founder and CTO of BSD Education, RefuGeek, and also is an educational advisor to the HKU faculty of education.

Work experience spans development, publishing, and digital marketing with experience in agencies, corporate, and hospitality markets. Over the last 12 years, Nickey has become an expert in developing and maintaining technology solutions, working with large-scale digital transformation projects, digital marketing, and the effective use of social media to drive business success and harness the power of data.

He specializes in developing teams in organizations to create sustainable and effective solutions themselves with a combination of consulting, training, and execution.

Will AI substitute teachers?

Written by guest Kevin Pereira, Blu Artificial Intelligence

As a part time lecturer at the University of Hong Kong (HKU) and Hong Kong University of Science and Technology (HKUST) and my consulting work at Blu Artificial Intelligence, I’m often asked how AI will impact education in the future. We tend to see popular media pushing the narrative that AI will take over teaching. I take the opposing view. AI will not replace teachers. If anything, AI will become a new tool in a teacher’s toolbox. AI will free teachers from administrative burdens, give them insights on student development, and let them focus on what they do best – helping students grow.

The truth is that today we are still quite far away from having robots and AI surpassing human beings. However, AI does tend to perform very well at repetitive, structured and well-defined tasks. Hence the belief that AI will take away our jobs tomorrow in my view is quite far-fetched. If anything, we should think about task automation rather job automation. Most jobs are made up of certain tasks, each of which may or may not be easily automatable. We can each look at our own job, consider the tasks & skillsets that are hard to automate, and then focus on those areas for professional development.

Let us take teaching as an example. With the recent restrictions from COVID-19 a lot of the classes I teach have moved fully online. I started teaching in 2018 so I did the class in-person the year before. When I compare online and in-person, I find that student interaction is much easier to facilitate in-person. There’s very little “please unmute your mic”, or “can you repeat, you’re cutting out” and my personal favourite, “can you HEAR me?” with the entire class responding “yes, we can”. It is also easier to get feedback, both verbal and non-verbal, from students. Are they laughing with you or at you? Given my jokes, perhaps I should leave this question unanswered for myself.

This confirmed to me that human interaction is an important part of education. When I ask students for feedback on the class, almost all preferred an in-person class. This also jives with my views on tasks that are hard to automate. Generally, anything requiring human interaction is a challenge for AI because people react differently to the same stimulus. The fact that A+B does not always equal C is a problem for AI. AI has started to address this with larger data sets and training, but it is not easy.

The big question then is what does this mean for teachers? I believe AI will augment our ability to be productive. This means that teachers will work with AI tools to create better student experiences. For example, AI can take over structured and repetitive administrative tasks. Grading is a prime example, and it brings back some memories for me. When I was little, my mom, who has been a teacher most of her career, used to get me to help her mark her students’ multiple-choice tests. After bribing me with my favourite candy, I would happily read off “A,C,D,E,B…” into the wee hours of the night.

Today, we have Scantron sheets for multiple choice grading. Soon, with an area of AI called Natural Language Processing (NLP), AI tools will be able to ‘read’ free form text responses and do the grading. I can testify that student handwriting standards have dropped, but if we give the AI enough data (handwriting samples) this can be addressed. Students could also type their responses, which negates the handwriting problem. On top of this, machine learning tools can construct ‘student profiles’ from grades to track their progression and identify development areas.

I know that many teachers, whether they admit or not, are reluctant to work with AI, and that is totally understandable. The AI isn’t perfect and will make mistakes. To expect otherwise is setting ourselves, and AI, up for failure. However, the potential to free up teachers to do what they do best is something that I feel needs to be explored.

If you’re scared about being substituted by AI, please take solace from students today who say they can’t wait to get back to the classroom. They need you. With that in mind, all I ask is that you stay open to AI augmentation and its potential to help you and your students.

If you found this interesting and would like to discuss further, feel free to reach out to me at kevin@blu.ltd.

About Kevin:
Kevin Pereira is currently a Managing Director covering Financial Services for Blu Artificial Intelligence, a consulting firm that specializes in Artificial Intelligence.

After growing up in Hong Kong, he started his career in Private Banking with Citi in New York working within both the Investments and Relationship Management areas. He then moved back to Hong Kong and joined Bank of New York Mellon’s Asset Management business, where he helped to start a new group that specialized in products tailored to High Net Worth Individuals.

Post business school, Kevin worked at a technology startup in Myanmar that was building out internet infrastructure which included fiber optic, cell towers and data centers. In this role, he spent half his time with the Venture Capital firm in Hong Kong and the rest on the ground with the portfolio company in Myanmar. Kevin is also a Visiting Lecturer at the University of Hong Kong (HKU), where he teaches a full- credit MBA course titled, “Artificial Intelligence for Business Leaders”. He also lectures at the Hong Kong University of Science and Technology (HKUST) where he teaches “Big Data in Finance”.

Kevin graduated cum laude with a Bachelor of Science in Economics with concentrations in Finance, Management and Marketing from the Wharton School at the University of Pennsylvania. He also has an MBA from INSEAD.