Artificial intelligence: Is it the future of test preparation?

This article is part cockyA series of articles about companies that use science and technology to solve problems in their industries.

Socrates taught Plato, and Plato taught Aristotle. Humanity has always known that the best education comes from an individual teacher. This is labor-intensive, expensive, and difficult to scale. This is what leads to the flawed, classroom-based education we have today: large class sizes and overworked and overwhelmed teachers. Educators are limited in their time and focus on either the brightest or the lowest performing students. Often, the wide quarterback is left to fend on his own.

Artificial intelligence could be an option for educators to solve these problems. Innovative technology based on computer code mimicking the brain’s networks of neurons can detect patterns in students’ performance and help teachers adjust strategies accordingly. “Artificial intelligence teachers” — software systems with which students interact online — promise to give each student individual attention, potentially reshaping education as we know it.

One of the few companies that are leading this transformation reedRed, as he pronounces it, is a Korean startup that he founded. YJ JangHe is a graduate of Berkeley’s Haas School of Business. Riiid is already a strong player in the Asian test prep market. Test of English for International CommunicationTOEIC (or TOEIC) is a test of proficiency in business English. Riiid is about enter. satThen there is the an actMarket preparation in the United States.

“Education is a complex field closely related to cognition, motivation, peer interaction, and so on,” Mr. Jang wrote in an email. “We draw insights from learning science, cognitive biology, data science, and other related areas of research for an iterative experimentation process that is both challenging and time-consuming — which is why there are so few players in the market.”

In the 1960s, the first computer-based lesson systems were created. They presented material in short clips, asked students questions and provided immediate feedback. These systems were costly and computers were not common at the time, so research institutes were the main beneficiaries.

Systems were first used in the 1970s and 1980s. Rule-based artificial intelligenceThen there’s the Cognitive theory. These methods guided students through each step of solving the problem and provided hints from specialized knowledge. But rule-based systems failed because they weren’t scalable — and programming vast domain expertise was expensive and tedious.

These systems were being set up at Berkeley by Mr. Jang when, Kanguk LeeProfessor at the University of Wisconsin-Madison, Zhang introduced him to deep-learning, an even more powerful form of artificial Intelligence in which algorithms learn by themselves, drawing from large amounts of data. Deep learning can be used to teach, with systems that learn content over time and student behavior.

He returned to Korea in 2014 to start Riiid. With a team of data scientists, he developed an AI set of algorithms that track student performance, predict results, and predict when students are likely to lose interest and drop out. Company published LeavesThis work was done at some of the top machine learning conferences around the globe.

Riiid launched an app to help you prepare for the TOEIC exam. Santa(Santa Claus collects data about children around the world. It quickly became a top-selling education app in Japan and Korea.

Riiid, a mobile app developer, has gathered data about student interactions and created what is now one of the most comprehensive public education datasets in existence, called EdNet. Riiid, however, has not been able to collect enough data to apply its AI system to a wider field of education.

“It is challenging to collect multimedia data that can be trained on AI in diverse learning environments,” Mr. Zhang wrote.

The company is currently focusing on the $300 million test prep market. Data collection is easy and the company has partnered in developing test prep apps with education companies around the world. Riiid collaborated with Casa Grande earlier this year to launch OE Saber – an app that helps students prepare for the Saber 11 college entrance exam in Colombia.

Riiid’s successes attracted a $175m Vision Fund II investment from SoftBank, which brought the company’s funding to around $250 million.

Riiid now offers an AI powered prep platform for the SAT/ACT college entrance exams. R.test, due to be released in January (prices not yet announced), projects standard test scores in a quarter the time it takes for a full mock test. Students will receive an analysis of their strengths and guidance on how they can improve by answering 30 questions. AI also allows them to select relevant practice questions. Riiid states that the intention is for students be able navigate the app and take the real test confidently knowing what their final results will look like.

“I really liked this because we could just use it at home instead of having a tutor,” said Esther Yee, a parent in Georgia who tried an early version of the platform. She liked the R test analysis. She said, “My tenth-grader would definitely benefit from it.”

Oscar Torres, a Chicago high school math teacher, tried the Riiid system. He liked Test R because it evaluates students’ knowledge in real-time without relying upon previous test scores. “With the development of artificial intelligence, I see it becoming a better and stronger resource for us,” he said. “We have to troubleshoot in real time as educators, and AI can help us tremendously.”

The company’s goal is much more than just test preparation. R.test is a tool to gather data and show the effectiveness of its algorithms. Riiid researchers continue creating new architectures, high-performance AI models, and other tools that can track student behavior and knowledge and help students select the best content.

“Our algorithms can predict student test scores with astonishing accuracy, a moving figure that acts as an incentive,” he stated. “The more students follow the algorithm’s recommendations, the higher their predicted score.”

Zhang believes that teaching won’t be based on intuition and guesswork, but on data. This may be the company’s greatest challenge. He said that collecting this data is a bottleneck due to privacy concerns making data collection in schools difficult. (Riiid states that its apps do not collect any personal information.

Riiid created a solution to address these concerns. EdSAFE AI AllianceA multi-sectoral coalition of companies, non-profits and education technology associations has been formed to develop norms and standards for the safe and responsible use and teaching of AI.

“The dream is for these algorithms to be integrated into a comprehensive system that can learn any subject to anyone, anywhere,” stated Mr. Zhang.