When OpenAI’s DALL.E 2 was released two weeks back, the AI tool’s ability to create images using sparse natural language instructions caused an online frenzy. Whatever its predecessor DALL.E could do, DALL.E 2 could do better. 

After the announcement, OpenAI’s CEO Sam Altman spoke about the potential upsides of DALL.E 2 and the general direction that AI was moving towards in his blog. According to Altman, the general idea that AI’s contributions would affect physical labour first, followed by cognitive labour and then eventually reach creative work has been reversed in reality. “It now looks like it’s going to go in the opposite order,” he noted. 

Source: Twitter

Altman speak

While the typical applications of AI like recognising and analysing text, voices, and images depend on finding similarities in data, creative processes are based on original thinking and do not necessarily follow a standard set of rules. The creative instinct is more complex and undefined than the analytical part of the brain, which is why it was initially believed to be an essential human quality. However, Altman seems convinced that AI will overcome its limitations faster than anticipated to fuel the creative scene.

In November last year, Altman tweeted out the demo for the code version where OpenAI’s Codex created a space game following natural language instructions. He mentioned, “People (like me!) with no artistic talent will get to experience the joy of creating in new ways, and actual artists will reach incredible new heights. The limiter will be the ideas you can come up with.” 

History of AI and art

The possibility that AI might become the next medium for art isn’t new. The first AI-generated portrait, ‘Portrait of Edmond Belamy‘ was sold at Christie’s in 2018 for USD 432,500. The painting was created by a group based out of Paris called ‘Obvious,’ which uses GANs or Generative Adversarial Networks to explore the connection between AI and art. GANs are built using RNNs or CNNs, where a generator and the discriminator work against each other to eventually fool the discriminator into thinking that the generated images are real-life. 

  Source: Obvious

Even prior to this, attempts were made to use AI to create art; when in 2016, a 3-D painting called ‘The Next Rembrandt‘ was made after analysing the artist’s portfolio using deep learning and facial recognition techniques. 

Film industries, too, are increasingly using AI-based storytelling in them. In 2016, IBM Watson was trained on trailers of 100 horror films to produce a six-minute trailer of a horror film called Morgan. In the same year, Benjamin, an AI system, was used to write the script for a short film, ‘Sunspring.’ The film contained certain irrelevant scenes. Later, in 2017, the system worked in tandem with film industry professionals to produce the sequel, ‘It’s No Game’, which was far more realistic and natural. This demonstrated that while AI could be used effectively used in art, it still required human assistance. 

Gaming and now films are also using 3D visuals to create more realistic and immersive imagery. For example, Vid2Vid is trained on videos of the landscapes in cities and then uses deep neural networks to churn out synthetic 3D surroundings for games. NVIDIA recently used a generative model called GameGAN that can be used by game developers to create fresh layouts automatically.

         Source: Twitter

Released in 2020, OpenAI’s enormous model GPT-3 can not only perform the basic natural language processing tasks like sentence completion, question-answering, translation and comprehension but also write poetry that needs far more complexity. The model has been trained on a huge corpus of 570 GB of text, which enables it to write essays, answer profound philosophical questions and write poems in any style of the user’s liking.  

Growing popularity but with exception

A recent survey conducted by Adobe showed that three-fourths of graphic artists based out of the UK, US, Japan and Germany would consider using AI tools to assist them in doing ‘non-creative’ tasks that are a part of the creative process like image search and editing. In fact, among the total respondents, there was only 11 per cent that wasn’t interested in learning about AI and ML at all. This indicates that artists are warming up to the idea of using more AI to aid them.

  Source: Adobe

However, despite how quickly AI has progressed to acquire creative skills, a study conducted by Crunchbase showed that companies still had the tendency to employ AI in gaming, advertising and other immersive applications more than finding creative usage. 

Jahan Singh Bakshi, digital and creative marketing strategist for films like Masaan, Lipstick Under my Burkha and Newton, spoke about how AI still lacked the contextual understanding that humans have. AI, he stated, can be used effectively for tasks such as generating thumbnails, but even then, it lacked the finesse that human involvement achieved. “There are times when I have seen thumbnails for films right next to each other with the exact same expressions,” Bakshi said.

“Art is born out of human experience. Unless an AI system has access to that level of depth in human emotion, pure human expression cannot be replaced,” he noted.