Are the AI Artists sowing chaos or creativity?
We have all heard of deepfakes, but what about deep learning? Both deepfakes and text-to-image generators are powered by a method of training AI called deep learning, which relies on artificial neural networks to replicate the function of neurons in the human brain. This has resulted in the rapid progression in AI’s ability to mimic the ways humans naturally speak and communicate. Image generators then allow users to create images from descriptions or upload and amend images. Hence, these image generation tools come with transformative potential, but also additional risks to society: worsening the spread of misinformation, reproducing biases and a host of creative concerns. Namely, can AI-generated images be considered genuine art?
The most well-known AI image generators, including Midjourney and OpenAI’s DALL-E, learn the underlying pattern related to the input and use that to generate similar content which enables photorealistic images to be produced. Consequently, Midjourney and DALL-E are now able to imitate well-known artists and genres, to generate credible photorealistic images of real-world situations and be considered to create ‘original’ artistic content. Hence, these platforms have been heralded as one of the most promising advances in AI, and led to experts predicting that as much as 90% of online content could be AI-generated.
Yet in the art world, AI’s ability to process vast amounts of art and spit out its own versions is significantly changing the role of the humans involved in the creative process. A notable example of this was the generative AI tool Midjourney being used to create the winning artwork in the digital art section of the Colorado State Fair, sparking accusations of cheating from other contestants. Although the artist was still initiating the creative process (via the text prompt), when it came to the images the role of the human has now been likened to that of a curator rather than a creator – merely selecting the best work from the huge amount of imagery the AI rapidly churns out.
That said, AI image creation has enabled dedicated art groups and digital galleries to appear. There is a very human desire to use AI to enrich and diversify the production of art and illustration. Yet, AI artwork has also been viewed as a form of theft and even a tool that risks sparking a decline in the overall creativity of humans if we become increasingly dependent on it for our creative ideas. Surely it is better for AI to handle repetitive tasks and for artists to use AI as just one of the options to gain creative insights rather than allowing AI to dominate the realm of creative tasks?
Beyond the art world, AI[1]generated imagery will likely influence various aspects of our online and offline lives. Most pressingly, AI-generated imagery has the power to shift our perceptions of what is real and true. This issue concerned many AI companies, including OpenAI, and consequently DALL-E prohibits images of celebrities or politicians to ensure users keep their text prompts ‘clean.’ OpenAI is thus attempting to reduce the flow of false or inflammatory images online. However, photorealistic faces are allowed to be generated by DALL-E and to be shared to social media, heightening the risk of physical-world consequences and dystopian outcomes.
One such concern is that image generators can output convincing images of riots and protesting – comparable to the January 6th Capitol Hill protests. It is all too easy to see how, when combined with effective use of social media algorithms, these images could be deployed by malign foreign actors or extremist groups to stir-up violence and unrest.
AI-generated images are also able to distort our present reality, or reinforce false historical narratives. Chinese technology company Baidu designed a text-to-image generator that expressly prohibits the creation of images of Tiananmen Square. Generative AI could very plausibly be employed to conceal other human rights atrocities such as those committed by Russia in Ukraine, or of Uyghur concentration camps in Xinjiang, China. AI is therefore being harnessed by authoritarian and oppressive regimes, reinforcing their censored version of history, to the detriment of society.
AI has resulted in shifting perceptions of human creativity. Some artists embrace the new tools, whilst others balk at them. But the greatest concern comes from AI’s reality-distorting abilities: specifically, how they can (and already are) being hijacked by authoritarian governments and used to sow political discord. AI-generated imagery is a double-edged sword, contemporaneously pushing boundaries and creating chaos.
Image: Generated with OpenAI’s Dall-E 2
