New machine learning model gives teachers access to high-quality creativity tests

Fostering creativity in schools is essential for an artificial intelligence (AI)-driven future. But while teachers embrace creativity as a vital 21st century skill, schools struggle to assess student performance in the absence of valid, reliable creativity tests.

Now, thanks to a new machine learning model developed by the University of South Australia, teachers have access to high quality, tailored creativity tests that can be assessed quickly and at low cost. Journal of Creativity Studies.

When applied to a current empirical creativity test, the Test of Creative Thinking – Drawing Production (TCT-DP), the new algorithm scores the test in 1 millisecond, as opposed to the standard 15-minute test scored by a human.

The development has the potential to save teachers thousands of work hours on their already packed schedules.

Lead researcher Professor David Cropley from UniSA says the algorithm represents a groundbreaking innovation for schools.

“Creativity is a vital skill for the next generation, and one that cannot be automated,” Prof Cropley said.

“But schools lack affordable and efficient tools to measure creativity, so students are either not tested or are given subjective assessments that are inconsistent and unreliable.”

“The TCT-DP test has long been recognised as the best tool for assessing creativity in school-age children, but it is expensive, time-consuming and labour-intensive, making it unfeasible for most schools to administer it.

“Our algorithm has changed this. Not only is the cost of running the algorithm more than 20 times lower, but the results are both faster and incredibly accurate.”

“For example, a manually marked test in a school of 1,000 students would cost around $25,000 and take around 10 weeks to receive test results. Using UniSA's algorithm, the same test can be administered for around $1,000 and results can be received within one to two days.”

“This will make the tests directly available to schools and teachers, providing an accurate and inexpensive means of assessing creativity.”

Co-researcher Dr Rebecca Marone from UniSA says the ability to test and measure creativity offers additional benefits to students who are often overlooked.

“Testing creativity opens up avenues beyond traditional intelligence tests,” Dr. Marrone said.

“Testing creativity can help us identify students who may have abilities that don't show up in traditional testing methods in schools. For example, a child who performs poorly on traditional IQ tests but is highly creative might easily be overlooked.”

“Fostering creativity can help protect low-achieving kids by training them in skills that are less susceptible to automation, and it can also help protect them from the effects of digital transformation.”

The algorithm is currently being developed as a desktop app that teachers can use in their classrooms. In advance of this, classroom teachers interested in using TCT-DP can contact the UniSA team to discuss their needs.