Advertisement
Rohit Sharma, Global Senior Vice President of Higher Education and Work Skills, ETS, told PTI in an interview about the Princeton-based agency’s plan for personalised assessments for the Test of English as a Foreign Language (TOEFL) and increased usage of Artificial Intelligence in the test.
“The ideal future of assessment is to provide every individual with a personalised assessment based on their background, making it as unbiased as possible. For example, if someone with an Indian background takes the test, is it the same as someone with a German background? Statistically, there’s no significant difference. However, it would be better if, when registering for the test, you could indicate your specific needs, like applying for a journalism programme, and the test could adapt to assess higher-level writing skills, for instance,” he said.
Sharma informed that ETS is also looking at personalisation based on background.
Related Articles
Advertisement
TOEFL is a standardised test to measure the English language ability of non-native speakers wishing to enrol in English-speaking universities.
The test is accepted by more than 12,000 institutions in more than 160 countries and is universally accepted in popular destinations such as the United States, Canada, Australia, and New Zealand, and by over 98 per cent of universities in the UK.
Earlier this year, ETS executed a series of changes in the 60-year-old test to create an optimal experience for those taking it.
Reducing the duration to less than two hours instead of three and allowing the candidates to be able to see their official score release date upon completion of the test, were among a few of the changes.
“There will be more changes in the near future… it may not always be to the test itself. We are incorporating more technology and AI, not just in the test but in everything around it,” Sharma said.
“In the next few months, we’ll release enhanced test prep materials, starting with TOEFL. We aim to help test-takers feel better prepared, focusing on areas they need to improve, he added.
We are also using AI for content generation and test security, especially for home editions of the test. We have different item types in the tests, such as sections focusing on comprehension, interpretation, and summarisation, Sharma said.
“Depending on where students are lacking, we aim to provide specific feedback on these item types, allowing them to work on very specific areas of improvement,” he said.
He, however, clarified that the use of AI by ETS is not new.
“We have been using AI for some time now. Initially, AI was largely used for automated scoring, but we’re moving towards using AI for content generation, which was more difficult and expensive before. Now, with advancements like ChatGPT and generative AI, the models are becoming much more accurate. We also use AI for test security,” Sharma said.
According to Sharma, AI helps detect and prevent cheating, which, although a small percentage in any industry, is still a concern.
“This is especially important for our home edition tests, where test-takers don’t go to a test centre but take the test at home. We are using AI to enhance our technology in this area as well. Eventually, we plan to use AI and the data we collect for personalisation,” he said.