Scientists identify key conditions to set up a creative ‘hot streak’ | Artificial intelligence (AI)3 min read
Whether it is the director Márta Mészáros or the artist Jackson Pollock, those in creative careers often experience a particular burst of success.
Now researchers have used artificial intelligence to reveal such “hot streaks” are commonly preceded by an experimental phase followed by a focus on one particular approach once the winning period has begun.
The director Peter Jackson’s career is, perhaps, a prime example: his hugely successful Lord of the Rings trilogy came after an eclectic range of movies such as the sci-fi comedy horror Bad Taste, the puppet film Meet the Feebles and the drama Heavenly Creatures.
The new work builds on a previous study by the researchers that suggested many creatives find themselves on a roll at some point in their career, although when exactly this happens appears to be random.
“About 90% of people have at least one hot streak” said Prof Dashun Wang of Northwestern University, who led the latest study. But, Wang said, it was important to understand why they happened.
“Then we can think about, how do we facilitate an individual to help them break through?” he said. “How do we first create an environment that will help the individual to achieve their full potential?”
Writing in the journal Nature Communications, Wang and colleagues report how they sought to investigate whether there was a common pattern behind hot streaks. To do so they looked at metrics of success such as the auction price of art works, IMDb ratings of films and citations of research papers to identify hot streaks for 2,128 artists, including Pollock and Frida Kahlo, 4,337 directors – including Mészáros and Jackson – and 20,040 scientists, including the Nobel laureates John B Fenn and Frances Arnold.
They then analysed how diverse the individuals’ work was at different points in their careers. This was assessed using an artificial intelligence system that was trained, in the case of art, to “recognise” different styles by features such as the brush strokes, shapes and objects in a piece, while in the case of film, it was trained to classify a director’s work based on plot and cast information. For science, the system identified different research topics based on the papers cited within a researcher’s publications.
The diversity before and during the hot streaks was then compared with the diversity at random points in the careers. The team found that for all three career types, work tended to be more diverse just before a true hot streak than expected from the randomly selected points.
However, once success had begun, individuals switched, sticking to a narrower than expected approach. That, the team says, suggests “that individuals become substantially more focused on what they work on, reflecting an exploitation strategy during hot streak”.
But the researchers found that neither exploring new approaches nor exploiting one were alone linked to hot streaks. Instead it is the combination that is important.
“There’s experimentation, and then there’s implementation based on what you have learned through experimentation,” said Wang.
Among other twists, the team found that scientists are more likely to try new things with small teams before a hot streak, but then work with large teams once a hot streak begins.
Wang said a key direction for future work is to look at how long an experimentation period tends to be.
Pamela Burnard, a professor of arts at the University of Cambridge, who was not involved in the study, welcomed the research.
“‘Trying something new’, ‘going for it’ and optimising ‘hot streaks’, are not new in our understanding of creative careers. What is new is the use of AI to study careers,” she said, adding that old models, for education, economies and creative careers were failing.