It would be illogical today to think that AI completely replaces human creativity. Having two such powerful “machines” and deleting one of them would be an absolute mistake. Instead, we should take advantage of 200% of the potential offered by both, an awesome combination impossible to replace.
Let’s talk about art, music, dance, writing, … “Being creative means being in love with life”, being able to generate new ideas or concepts spontaneously.
Does AI take place in these fields? Or should we reserve it only for humans?
Portrait of Edmond de Belamy, an AI painting which has been sold for $432,000. Its creator? An algorithm, specifically a generative adversarial network (GAN) that was fed a data set of 15,000 portraits covering six centuries to inform its output.
What if we talk about Ai-Da? “The Worlds First Ultra-Realistic Humanoid AI Robot Artist”. Equipped with eye cameras and robotic arms, it is capable of painting a picture or speaking thanks to a simulation model called Human-In-The-Loop or HITL.
Progress in AI music field has rapidly accelerated in the past few years.
If we go back in the past, in 1951 Alan Turing was the first human to record computer-generated music using a machine that filled almost an entire floor of the lab.
One of the most used techniques today is reinforcement learning, an algorithm that learns the features and patterns found in songs to replicate or simulate new ones of a certain music genre. Google’s Magenta Project would be an example, an open-source platform that produces songs written and generated thanks to AI, or also IBM’s Sony Watson Flow Machine solution. In fact, Alex Da kid, Grammy nominated producer, relied on this tool in his “creative” process.
Have you heard about AIVA? “The AI composing emotional soundtrack music” comes to be known as the “Creative assistant for Creative People”, it allows you to create music based on different styles or influences: modern cinematic, electronic, pop, rock, fantasy, among others.
At Peltarion, an artificial intelligence (AI) expert and dance choreographer was a clear example of the application of AI in dance. The team managed to create a solution not only based on learning a certain style of dance, but focused on creating its own choreography. How did it work? It consisted on a recurring neural network called “chor-rnn” that fed from +13.5 million different positions from 5 hours of contemporary dance. In just 48 hours they achieved their first results.
NVIDIA was also capable of creating new dance moves according to the style of a song, following the beats of the music. They used a MM-GAN (generative adversarial network) collecting different videos from Ballet, Zumba and Hip-Hop. Thanks to this solution, they managed to generate 361,000 clips with a total of 71 hours of new dance steps.
Within our last field of “creativity”, we can find what we call Natural Language Generation (NLG) which consists of creating “narrative writing” from data. For example, the american Associated Press is currently using AI to write thousands of sports reports.
New websites have also emerged generating “unique” text from one single headline to create a whole complete draft of a new article. The process is so simple: Choose a topic or headline, wait a minute and take what the automatic article writer drafted. Incredible, don’t you think?