How Scientists Use Machine Learning To Measure Creativity With a Simple Word Test

The beauty of creativity is that it supersedes the limitations of time and space. Whether it is pen to paper, or water colour to pad — creativity is incredibly hard to define and pinpoint.

Curiously, I can pride myself on balancing data analytics with compassion, but I can draw and paint if the occasion presents itself.

People often get surprised at the two facets of me, especially if some people are more acquainted with my scholarly and data-driven side, and others are more acquainted with my creative writing and compassionate side.

Measuring Creativity Through Deep Learning

Given the subjective nature of creativity, it is very difficult to measure this construct quantitatively. Yet, researchers published a study where they have created a four-minute test that examines at least one facet of creativity: divergent thinking.

Using deep learning algorithms, measurements between two seemingly related and unrelated concepts can be made. For example, similar to a human brain, imagine a large neural network where many concepts are floating around.

These concepts all have complicated relationships to one another, meaning that some semblance of pattern and prediction can be made over a period of time. This is particularly helpful if scientists want large aggregates of meaningful data in a short span of time.

Simply put, these algorithms capture complexity better than a regular person ever could— especially with our biases, limitations, and beliefs.

Defining and Exploring Divergent Thinking As the University of Washington suggested, divergent thinking pertains to our ability to come up with many different ideas about a single topic — within a short span of time.

It may or may not involve breaking down a topic into various pieces for additional insights and coherence. This also requires asking a lot of questions for further brainstorming. These include questions surrounding our:

Daily use of time Perception of knowledge Personal interests Deep-seated values and beliefs To give coherence to divergent thinking, you need to narrow concepts down, such as identifying:

Root causes and effects Descriptions of various concepts Purposes and intent Categories, relationships, and biases To optimize divergent thinking, you will need to further brainstorm, physically map out your ideas, research, engage in journal writing, and freestyle for as long as you need to.

There’s nothing wrong with a little logic when it comes to creativity — they often go hand-in-hand, especially when niche problems require creative solutions.

Explaining the Study in Layman’s Terms Created by scientists from McGill University, Harvard University and the University of Melbourne, the Divergent Association Task (DAT) is a game that can be played by many people, regardless of culture and age.

You see, the study examined over 9000 people, ranging from the ages of 7 to 70. The participants lived in 98 countries, to add to the cross-cultural universality and credibility.

For the first half of the DAT, participants did something called the Alternative Uses Task (AUT) where individuals were asked to come up with new ways to use an object. Then, they were judged based on the difference between the established use of the object, and the answer they had provided.

For the second half, the Bridge-the-Associative Gap Test (BAGT) was used to provide participants two seemingly unrelated (or related) words. They were then asked to provide a third word that had linked the first two words together. For example, words like “critters” and “reading” could be linked together through words like “bookworm”. Results of the Study

Both parts of the DAT were judged by deep learning algorithms, instead of people. As mentioned earlier, these automated judges examined the relationships between concepts through divergent thinking. This is because deep learning increases the objectivity of the test and can analyze many people in a short span of time.

Based on the results, the demographics mattered very little, but younger people, including women, had a slight statistical advantage in creativity. This means that the test is cross-culturally sensitive, and could be just as reliable as measuring one’s verbal creativity, or even one’s creative intelligence. Implications for the Greater World This study impacts the greater world, as creativity is limitless and machine learning is able to map out complexities beyond the limits of human cognition and biases. The test was able to take into account the hard labour, pressure, and work that went into a project — and not just the end product.

For example, throughout time and history, people engaged in various bouts of creativity and often went beyond the established norms of society — at great cost. They were only celebrated once their published art hit the markets, but very few people may not realize the creative energies that flowed during the creation of that masterpiece.