Online math lessons service uses artificial intelligence to help boost students' skills and confidence

Online math lessons service uses artificial intelligence to help boost students’ skills and confidence

Like many students around the world, Ethan, 14, in Corley, UK, has been struggling to keep up with math at school after more than a year of turmoil linked to COVID-19. In June 2021, her parents enrolled her in a summer program presented by Eedi, the online mathematics teaching service.

“Just dealing with the lockdown, she just didn’t have enough of a really good background,” her mom, Ariana, said. “She missed most of year seven in math, then year eight. So, we thought, ‘Let’s give it a try, see where she needs a little help.'” “

Newly enrolled Eedi students are required to take a dynamic test of 10 diagnostic multiple-choice questions that the service uses to learn where students struggle the most in math. This information allows the service to put students on a learning path to overcome those specific obstacles or misconceptions.

“We ask them a question roughly based on their age group and then say, ‘Well, what’s the best next question to ask them based on their previous answer?'” explained Iris Hulls, Eedi’s COO. “We learn as much about them as we can to predict growth or comfort issues for them.”

Dynamic testing is powered by advanced AI by researchers at the Microsoft Research Lab in CambridgeUK, specializes in machine learning algorithms that help people make decisions.

The AI ​​uses each answer to predict the probability that a student will correctly answer all of the thousands of other possible next questions and then balances these probabilities to decide which question to ask next to identify knowledge gaps.

The information obtained from the test is similar to what the teacher might learn from a one-on-one conversation with the student, explaining Cheng Changa Microsoft Principal Investigator in the lab who led the development of a machine learning model that supports dynamic testing of the Eedi.

“If a student doesn’t know 3 times 7,” Zhang said, “we might want to subtract 1 plus 1.” “We want to adapt the test based on the previous answer.”

Once students’ misconceptions are identified, the Eedi platform inserts students into an educational path that helps them overcome their misconceptions and achieve better math performance in school.

Ethan was put on a course that included a review of the topics covered in Year 8 and prepared for success in Year 9, including engineering.

“It is very good to find your weaknesses and strengths and to be able to understand why you are not satisfied in that particular area,” said Ethan. “You can realize, ‘I’ve been doing this wrong for a long time. “

Ethan, 14, in Corley, UK, gained confidence in math through lessons on Eedi, an online learning service that uses artificial intelligence developed by Microsoft. Photo by Jonathan Banks.

Good questions, good data

Microsoft success Next best question template Depends on the data used in training, Zhang noted. In Eddie’s case, these are thousands of high-quality, screened diagnostic questions that have been specifically developed to help teachers identify students’ misconceptions about math topics.

“Our technology is just an optimizer that makes this high-quality data give more insights,” Zhang said.

Diagnostic questions are well-researched multiple-choice questions that have one right answer and three wrong answers, with each wrong answer designed to reveal a specific misconception.

“Mathematics fits well with this kind of multiple-choice assessment because there is often a right answer and these wrong answers; it is much less subjective than some subjects in the humanities.” Craig BartonCo-founder of Eedi and Director of Education for the company.

Barton held on to the power of diagnostic questions when, as a math teacher, he attended a course on formative assessments and learned that well-formulated wrong answers can provide insight into why a student is struggling.

“In the past, kids always got things right, which is a good thing,” he said, “or they got things wrong, and then I had to start doing detective work to see where they went wrong.” “That’s fine if you’re working person to person, but if you have 30 kids in a class, it’s probably going to take a long time.”

Good diagnostic questions should be clear and unambiguous, check one thing, be answerable in 20 seconds, associate every wrong answer with a misconception, and ensure that the student is unable to answer them correctly while there is a major misconception, Barton said.

“This idea that kids can’t get it right with a major misconception is the hardest to keep in mind, but it’s probably the most important,” he said.

For example, consider the question: “Which of the following is a multiple of 6? – A: 20, B: 62, C: 24, D: 26.”

According to Barton, this is an ostensibly decent question. This is because students may think that “multiple” means that “6” is the first number (B) or the last number (D), or the student may have difficulty with their multiplication tables and choose A. The correct answer is C: 24.

“But the main drawback with this question is that if you don’t know the difference between a factor and a multiplier, you can get this question right, whereas experience will tell us that the biggest misunderstanding that students have with multipliers is that they confuse it with factors,” he said.

A better question to ask is, “Which of these is a multiple of 15? – A: 1, B: 5, C: 60 or D: 55.” That’s because possible answers include factors and multiples. The correct answer is C: 60. A student who mixes factors and multiples can choose A: 1 or B: 5, and a student who needs work on multiplication may choose D: 55.

“When you write these things down, you really have to think, ‘What are all the different ways kids can go wrong and how am I going to pick up these three wrong answers?'” Barton explained. “

Screenshot of an online math test asking for a five-digit average with four answer choices
In this diagnostic question, the correct answer is “B:4”. Students who answered “A: 20” took the first step to find the mean, totaling the numbers. C:3 represents the confusion between the concepts of median and median. “D:2” is a combination of conceptualizing and averaging.

Teacher tools for online teacher

After the workshop, Barton came home and wrote about 50 diagnostic questions and tested them on the students in his class. They worked.

Barton is also a maths author and podcaster with thousands of followers on social media. He used his influence to spread the word on diagnostic questions and collaborated with Eedi co-founder Simon Woodhead to build an online database of thousands of diagnostic questions for teachers to access in order to plan their lessons.

“Then I thought, ‘Wait a minute, we can do something a little better than this,’” Barton said. “Imagine if kids could answer questions online and we could capture that data, and then, before you know it, we have insights into specific areas of suffering.” where the students are.”

The website went viral and attracted investors as well as the interest of Hulls, who and his colleagues were exploring options for using the data to extend the benefits of math lessons and make them more accessible to more families. Team form Eddie. A consultant introduced them to Zhang and her team’s research on the Next Best Question algorithm, which aims to speed up the decision-making process by collecting and analyzing relevant personal information.

At the time, Microsoft researchers were working on healthcare scenarios, using artificial intelligence to help clinicians make more efficient decisions about which tests to order to diagnose patients’ illnesses.

For example, if a patient walks into the emergency room with an arm injury, the doctor will ask a series of questions that lead to an X-ray, such as “How did you hurt your arm?” and “Can you move your fingers?” Instead of “Do you have a cold?” Because the answer will reveal information relevant to the treatment of that patient. The Next Best Question algorithm automates this information gathering process.

The consultant thought the model would work well with the Eedi dataset for diagnostic questions, automating the process of gathering information that the teacher could glean from a one-on-one conversation with the student.

We knew we had collected a lot of data. We wanted to do smarter things with our data; “We wanted to be able to predict the misconceptions students might have before they answered the questions,” said Woodhead, Eddie’s chief data scientist.

The Eedi team worked with Microsoft researchers to train the model on their diagnostic questions to identify where students need the most math support.

Woodhead noted that the model works without collecting any personally identifiable information from students.

“He doesn’t need to know a name. He doesn’t need to know an email address. She’s looking at patterns,” he said.

From this information, the system can determine the best lessons for students to learn in Eedi. Without this guidance, students tend to rely on strategies they already use in school, which is not the right starting point for the majority of students looking for a private tutor, according to Holz.

“It really helps guide the kids and their families back home to know where to start,” she said.


#Online #math #lessons #service #artificial #intelligence #boost #students #skills #confidence

Leave a Comment

Your email address will not be published. Required fields are marked *