An artificial intelligence-based TOEFL grader developed by a Korean startup is in the spotlight for fast evaluation, score prediction and personalized tips to get good marks as well as saving costs for Koreans who are behind the bulk of the revenue for the standardized test for foreigners aspiring to study in the United States.

The AI tutor called TOEFL Bank is engineered to provide users with the shortest path to their target score, Song Da-hoon, CEO of Data Bank, the operator of the system, told Maeil Business Newspaper on Wednesday, adding AI-based automated evaluation can contribute to reducing private tutoring expenses and closing the regional gap in English education.

Unlike conventional automated scoring engines which rely primarily on rule-based matching, TOEFL Bank AI focuses on natural language understanding based on deep-learning algorithms, according to the company. TOEFL Bank AI can provide consistent and accurate assessment because it is constantly learning from 10 million massive user data records.

TOEFL Bank has accumulated 11,456,434 user learning data from more than 400,000 cumulative users to date. TOEFL Bank`s grading system gets more sophisticated over time thanks to thousands of data from around the world every day, Song explained.

Based on this strength, TOEFL Bank saw the biggest increase in the number of users last year. The cumulative number of visitors spiked to 400,844 persons in April this year from 87,016 a year ago. The number of subscribers increased nearly 20 times from 5,517 to 115,273 in the same period.

In April, Data Bank attracted 3 billion won ($27 million) from investors. TOEFL Bank plans to introduce a paid service from the third quarter of this year, with a swift to a premium service after a certain period of free services.

By Chung Ji-sung, Han Jae-beom and Minu Kim

[ⓒ Pulse by Maeil Business Newspaper & mk.co.kr, All rights reserved]