Four AI tools that facilitate the learning process

Learning is never a straightforward endeavour. Every learner is unique and learns at their own pace, something that educational institutions are slowly beginning to address. To help students learn at their own pace and, more importantly, master the curriculum they are going over, educational institutions are now increasingly turning to online platforms that utilise the power of machine learning and artificial intelligence to offer students a personalised path to knowledge and mastery of certain subjects. With the right mindset and approach, these platforms promise to improve student performance and knowledge retention.

Quantifying knowledge retention

One such platform is ALEKS (Assessment and Learning in Knowledge Spaces), developed at the University of California by a team of cognitive scientists, software engineers, mathematicians, and educators. Based on the concept of Knowledge Space Theory, a framework for systematically quantifying knowledge, ALEKS is an online platform that uses adaptive machine learning to precisely determine each student’s knowledge level and lead them through a personalised learning experience. Used primarily for mathematics, the platform avoids multiple-choice questions and has students actually fill in formulas and solutions to problems. The questions used for assessing student knowledge are both adaptive and comprehensive, spawning, for example, multiple questions from the same topic if a student is struggling to provide good answers so as to more fully ascertain if there are any knowledge gaps. Since it’s online-based, ALEKS can be used both in school or at home. In practice, it is mostly used at home as a supplementary tool for struggling students. The platform has modules for all levels of education up to tertiary, including independent learners.

After the initial assessment, students are periodically tested to ensure progress in learning. ALEKS’ approach to instruction is assessment-based and each question is asked in relation to the student’s answers to previous questions. Assessment results are used to point students to study materials that will help them overcome knowledge gaps, but also highlight what new topics they are ready to learn. They are also always framed within the context of previously specified educational goals or standards. The results are presented in the form of a pie chart, where each coloured slice represents the curriculum area and, if clicked on, can list all the topics that the students are ready to cover next.

In the Learning mode, which comes after the assessment is finished, students can do practice problems, look at explanations of concepts and problems, get feedback from the system, and monitor their own progress. If a student is trying to solve a mathematical problem using the wrong methods, ALEKS will provide hints or propose that they work on a different problem. The student can also ask for a step-by-step explanation. The Teacher mode allows educators to monitor student progress and change learning parameters and expectations. They can get a report on every student, which includes assessment results, progress in the learning mode, and the total time spent on the platform. The platform also sends an email notification to the teacher if a student is exhibiting behaviours that require intervention, such as if they are struggling, aren’t showing progress, or are procrastinating excessively.

A rich, integrated learning experience for everyone

Maths and science subjects tend to be especially challenging for students. One reason for this is that many education systems provide children with a prescribed learning path without taking their individual characteristics and preferences into account. As a result, some students progress faster while others lag behind. One way to tackle this problem is to personalise the learning experience for each student’s individual learning needs with the help of AI. South Africa-based tech firm Siyavula Education has developed an online learning platform that draws on the latest research into motivation, sequencing, cognitive science, and machine learning to optimise practice sessions in mathematics and physical science subjects for secondary school students. It also provides students with free access to a comprehensive catalogue of openly-licensed science textbooks approved by South Africa’s Department of Basic Education. “Our mission is to champion a love of maths and science by creating rich, integrated learning experiences for people everywhere through the use of science and technology,” says Mark Horner, the company’s CEO.

Siyavula aims to make every student’s learning journey personalised through the use of a machine learning engine. Its features come to the fore in the Siyavula Practice module, where each practice session is tailored to each individual’s education level and set of skills. For example, if practising maths, students are given problems to solve and the platform marks their answers, after which it provides them with a more complex problem to solve. The algorithm behind Siyavula Practice makes sure that exercises are ‘interleaved’ to make sure that students are practising more than one skill at a time. Students can track their progress through the Learner’s Dashboard, where they can see which concepts they have already mastered and which ones require more work. At the same time, teachers have their own dashboards for tracking cumulative progress of their students, where they can see which areas need more instruction and how much time students are spending on a certain topic.

Currently, Siyavula reaches 18 per cent of the South African youth, who are most economically disadvantaged. The goal is, of course, to positively impact as many students as possible, and that is why Siyavula has embraced open web standards to ensure maximum reach. The platform supports everything from a smartphone to a PC and does not require servers. The designers behind Siyavula have made an effort to make Siyavula’s textbooks available even through the most basic mobile phone.

AI takes over test prep

AI can also help students prepare for taking language exams. English has become the world’s lingua franca, particularly in international business settings. Therefore, any student who wishes to improve their career options needs an advanced knowledge of English and high scores in standardised English tests. In the past, specialised language schools would offer test preparation, but now there are AI-powered apps that promise the same, if not better, quality of test prep. Korean educational technology company Riid offers AI-powered learning software for English learners. Youngduck Choi, the AI tech lead at Riid, explains that Riid applies state-of-the-art AI models to advanced tasks such as dropout prediction, score prediction, and optimum learning path generation. Furthermore, their technology uses datasets that consist of students’ learning histories, lecture-watching, and question-solving logs, which ensures that each student receives educational content that fits their needs and abilities.

Available on Android, iOS, and via browser, Riid’s Santa TOEIC is a learning platform aimed at English learners who are preparing to take the TOEIC (Test of English for International Communication). Once a user signs up for the platform, they take a short placement test that consists of no more than 10 questions. The answers are used to create a customised learning path that will ensure the learner improves their score as quickly as possible. But the user analysis doesn’t stop there – even as the learner is using the app, Santa TOEIC analyses their results in real time and offers lectures every time it identifies an area for improvement.

Riid has clearly discovered a winning formula, considering that English language learners who are using Santa TOEIC achieve an increase of around 130 points on average after spending 20 hours using the platform. Currently, there are over 1.1 million users doing their test prep on the Santa TOEIC platform, which is available to both Korean and Japanese learners. Despite the success of Riid’s virtual platform, Choi believes that AI learning models can’t wholly replace the entire educational experience, which includes human communication and non-quantifiable qualities. But he does expect such AI tutors to take the place of human teachers when it comes to learning with measurable goals, such as in standardised test preparation.

A reliable source of factual answers

Another area where AI can prove rather helpful is in providing factual answers to all sorts of questions. As a student, Stephen Wolfram was passionate about physics. But despite his gifts as a young scientist, he had always struggled with maths. Understanding concepts was simple enough, but doing the actual calculations appeared to be an insurmountable challenge. That is why he turned his attention to computers, whose enormous computing powers he wanted to use for calculations, while keeping his attention on the larger picture. In 2009, Wolfram released Wolfram Alpha, a powerful search engine for computing answers and generating reports from the vast repository of  human knowledge – an ambitious project, to say the least. In the words of its creator, Wolfram Alpha aims “to take all areas of systematic knowledge and make them computable.”

While Wolfram Alpha has a clear bias towards maths and sciences, its engine covers topics from everyday life, to history, the arts, and politics. It uses natural language processing to understand the questions it’s being asked and, drawing from datasets based on primary sources, it generates a factual answer or report. If asked to solve an equation, it will break down the solution step by step, which means it can be (and is being) used as a homework assistant. It differs from search engines in that it does not simply point to a list of possible sources but, rather, it presents the most relevant and factually accurate answer it can, in a systematic fashion. Obviously, this also means it’s limited by data and it cannot answer narrative questions (for example: What is the difference between realism and romanticism?).

Recently, Barnes & Noble Education (BNED) signed a deal with Wolfram Alpha to develop a maths solver as part of its bartleby homework suite. The maths solver will enable students to input equations on a mobile or web platform and receive a step-by-step explanation in real time. “Working with Wolfram Alpha to develop a maths solver is a natural expansion of our learning ecosystem, combining Wolfram’s computational knowledge and academic credibility with bartleby’s high-value learning solutions,” said Michael P Huseby, CEO of BNED. Wolfram Alpha also has some 50 course assistant mobile apps for students. These are specialised apps that help students understand the problem, rather than simply getting an answer. Some of the courses they cover include: physics, music theory, geography, astronomy, and general chemistry.

AI-powered tools have taken on an increasingly prominent role in educational settings in recent years. Not only are they a neat way for educators to ensure their students are studying outside of school, but they can be utilised in a variety of teaching models, from the flipped classroom to the hybrid learning model. And having a system in place that pinpoints areas in which a certain student is struggling can be a great help to overworked teachers, allowing them to provide timely and relevant interventions that significantly enhance the student learning experience.