Part of:

Can Creativity Be Implemented in AI?

Why Trust Techopedia

Is creativity strictly a human capability, or is it possible for artificial intelligence to actually be creative?

Artificial intelligence (AI) takes the power of computing systems to a different level. It is amazing to even think that a computing system can emulate human beings. There are many fantastic examples of AI in various areas of our lives. That said, computing systems are still considered limited in their capabilities because they cannot think creatively like human beings. While AI can process and analyze complex data, it still does not have much prowess in areas that involve abstract, nonlinear and creative thinking. For example, it is extremely difficult for a computing system come up with a groundbreaking scientific theory like that of relativity. Can AI overcome this limitation? AI is being enriched regularly, but nothing much has been done so far to catapult AI to the next level.

What Is AI?

AI is an area of computer science that studies intelligence in computing systems. Though it may sound a little strange, AI enables computers – to an extent – to think, react and work like human beings. Intelligent computers can do many different things like planning, speech recognition and problem solving.

Exploits of AI

As already stated, the exploits of AI are too numerous to be chronicled here. Still, some of the most notable achievements are briefly described below:

  • AlphaGo, the AI software from Google, beat Lee Sedol, the world champion, in the extremely complex Chinese game of Go. Go is like chess in terms of making moves, but unlike in chess, it is impossible to calculate all possible moves because the number of possible moves in Go are nearly infinite.
  • Google’s AI software wrote poetry. The software was fed more than 11,000 poems. Based on the data from those poems, the software composed new poems.
  • At Tufts University, an AI software developed a scientific theory on the regeneration of flatworms. The topic had been a mystery for 120 years.

Can AI Really Become Creative?

Despite its achievements, it is hard to believe that AI could actually become creative – at least not anytime soon. Think of the examples described above. The common tendency in each of them has been the dependence on data – huge volumes of it. The machines first need to process and analyze data given to them before doing anything novel. All that AI can most likely do is to find a new pattern based on the many patterns already given to it. That goes against the basic principles of creativity. The human mind cannot store or process such huge volumes of data, but that does not prevent it from creating something outrageously novel.

One area that is extremely difficult for AI to dabble in is pure arts. According to Michael Osborne, associate professor in machine learning, University of Oxford, it is extremely difficult to teach algorithms to produce art like that of human beings. It is possible to train algorithms to churn out pieces of artwork in large volumes, but it is difficult to teach them the difference between quality and poor art. A survey conducted by The Guardian, a reputed newspaper in the U.K., found that in the U.K. and the U.S., almost 90 percent of artistic jobs cannot be automated.

From the opinions given by eminent figures in the field, it does not seem that AI can become truly creative anytime soon. Let us review a couple of these opinions.


David Cope, Composer, Author and Professor Emeritus of Music at the University of California, Santa Cruz

Professor Cope has been trying to have computers write novels for a long time and has achieved some success. Computers are now able to write short stories, but questions arise over the quality of the stories. According to Professor Cope, there may soon be a time when AI can churn out 10,000 words in a mere 30 minutes. But can such stories give joy and value to their readers? Probably not. The short stories written by the machines are related to one another, which provides data for the computers to analyze. The basic elements missing here are creativity and novelty. Computing systems rely on previous data input even in writing short stories.

Maria Teresa Llano Rodriguez, Research Associate, Computational Creativity Group, Goldsmiths University

According to Rodriguez, AI’s creativity is constrained because of the type of data provided. She elaborates that the data quality, variety and volume are important factors in enabling AI to become creative. There has been an overall failure in providing such data to AI systems. Though no doubts are cast on the ability of AI in this case, it still holds true that AI is dependent on data quality.

How Can AI Become Creative?

AI, it seems, can improve, but is unlikely to match the human mind. There are certain areas, however, where AI can claim to achieve total mastery, such as driverless cars and vehicle manufacturing. In fact, such industries have already been going through large-scale automation. To improve the capabilities of the algorithms, they need to be constantly supplied with huge, updated and varied volumes of data so that the machines can adapt and learn. Based on the learning, it can find novel things. But areas like psychology, medicine and art remain unconquered by AI.

Let us take the case of the horror movie “Morgan” where the IBM Watson cognitive platform played a prominent role. Basically, Watson reduced the huge effort and time that humans would normally spend on making the trailer of this horror flick. With pure human effort, it would have taken significantly longer. Watson was fed numerous trailers of horror thrillers. Watson analyzed the visuals, sound and composition of the trailers and selected the most appropriate clips. You can say that Watson’s efforts were close to that of a human brain – if not equal – because it could analyze and identify the appropriate clips.

This is how AI works: imitation. Basically, AI mimics the data it is given as input. You feed it large volumes of data, then AI processes and analyzes the data and finds new patterns, which some people call creativity. According to Jason Toy, CEO of Somatic, a start-up that develops deep learning applications, AI works on the deep learning principle. “If you feed it thousands of paintings and pictures, suddenly you have this mathematical system where you can tweak the parameters or the vectors and get brand new creative things like what it was trained on.”

The people who believe AI can be creative believe so because its achievements, unthinkable in the past, have increased by leaps and bounds. For example, no one believed that a computer could distinguish between what is and what is not cancer, but Watson is currently working on that very task. Basically, such people are relying on the evolutionary trends of AI. Indeed, AI has achieved a lot in a short period of time. But what is ignored is that AI has done different things with a common approach – deep learning and mimicking data. But creativity demands independence.


There are three implications of the AI becoming creative:

  • One, it can potentially replace human beings in certain domains. It has already been causing massive disruption in such domains.
  • Two, AI and human beings will complement each other in certain domains. For example, repetitive tasks will be left to AI while the more creative jobs are done by the humans.
  • Lastly, certain domains will remain almost fully unconquered by AI.


AI is projected by many as something catastrophic for mankind because it is going to take away jobs. While that may be partly true, AI can potentially unlock a great future for us by actually forcing greater innovations. The way forward is coordination between humans and AI.


Related Reading

Related Terms

Kaushik Pal
Technology writer
Kaushik Pal
Technology writer

Kaushik is a technical architect and software consultant with over 23 years of experience in software analysis, development, architecture, design, testing and training. He has an interest in new technologies and areas of innovation. He focuses on web architecture, web technologies, Java/J2EE, open source software, WebRTC, big data and semantic technologies. He has demonstrated expertise in requirements analysis, architectural design and implementation, technical use cases and software development. His experience has covered various industries such as insurance, banking, airlines, shipping, document management and product development, etc. He has worked on a wide range of technologies ranging from large scale (IBM…