As artificial intelligence (AI) continues to develop, it has begun entering the creative space — one we had assumed would be the last one to be addressed by AI. But instead, we now have AI that can write, paint, create music, and so much more. In fact, most of what you see on this page was made by an AI model. The visual for this article was created using MidJourney AI, while ChatGPT wrote parts of this article.

In the past couple of months, ChatGPT has become the fastest-growing web app ever, already surpassing 100 million users. Its incredible writing ability combined with its ease of use has made it exceedingly popular in almost every industry where written content is needed. From writing tedious emails to full-blown newspaper articles, ChatGPT has excelled nearly everywhere, and this is still a research model that is growing and learning for future iterations.

To see how convincingly human these models can be, let’s play a game. In this article, two sections have been written using a ChatGPT model trained to mimic my writing style. See if you can figure out which paragraphs were written by me, and which were written by a bot.

Who knows, maybe you’ve already read one that wasn’t written by a human.

How it works

AI has been making rapid advancements in recent years, leading to increased abilities in content creation, and with increasing computational power, AI models are becoming better at producing written, visual, and audio content that resembles human content.

But AI models develop creativity in a very distinct manner from humans.

While humans consider creativity an innate process that inspires us, creativity in AI works by training models on large datasets and then generating content based on the patterns learned from that data. For example, DALL-E, which is capable of generating highly detailed images, is trained on thousands of images, while ChatGPT, which can generate human-like text based on a given prompt, is trained on text.

This process enables the models to learn how different elements of an image or text relate to each other and how they combine to form a coherent picture.

Once the models are trained, they can generate new images or text by combining the learned relationships and patterns in new and creative ways. For example, when generating an image, the model might start with a basic shape and then add elements like colour, texture, and details based on what it has learned from its training data. Similarly, when generating text, the model might start with a prompt and then generate text based on what it has learned about the relationships between words and phrases.

It’s important to note that AI models like DALL-E and ChatGPT don’t truly understand the images or text they generate. They don’t have a concept of the meaning of the objects or words they generate. Rather, they are simply combining patterns and relationships in the data they have seen in a new way. This means that while the results can be impressive, they don’t mean anything to the model itself.

But despite the lack of understanding, AI models still have an impact on audiences. Regardless of its origin, content that engages, informs, or inspires its audience is what truly matters. In the end, the goal is to use AI and human creativity in a complementary fashion, allowing for the creation and dissemination of even more powerful and impactful content.

A creative prodigy

The uses and applications of AI-generated content will be vast. From generating articles for a newspaper or personal blog to developing orchestral works, there is no limit to the possibilities of AI. Its only limit will be human creativity.

One of the fastest-growing uses of creative AI is in the entertainment industry. AI can generate music, write screenplays, and create visuals that rival the works of human artists. AI-based platforms like Amper Music and Artificial Intelligence Virtual Artist are even allowing musicians to create new samples and tracks based on a few simple parameters like mood, length of track, and genre they want. 

Along with AI that can generate screenplays, the development of ‘deepfake’ technology and voice mimicry is making it so that actors would not even need to be present physically while acting out a scene. This notion has also caused controversy, particularly around the idea of using deepfakes of actors who have passed away, almost reviving them digitally.

Continued development of creative AI is also reducing the need for human-human interaction, particularly in the service and education industry. AI-powered tutors are becoming a notion now, with chatbots like Cognii creating assessments and grading systems while others like Replika develop interactive educational games to teach classes.

Alongside the generation of content, AI can also be used to make existing content more accessible to humans. It can be used to automatically generate captions and transcripts, as well as translate videos in real time for individuals to watch or listen to.

A trial for humanity

Creative AI can have many benefits, but it also poses potential risks and challenges that need to be addressed, particularly its potential for misuse.

One such case is where AI can be used to generate fake or misleading information, such as deepfake videos that can manipulate people’s opinions or spread false narratives. 

Deepfake technology has been a growing concern in recent years as it becomes easier to manipulate audio and video content. With AI’s ability to generate realistic images and sounds, deepfakes can be used beyond the film industry, to manipulate public opinion or spread false information. In some cases, deepfakes can even be used to harass or defame individuals, particularly in the case of political figures. This malicious use of AI in deepfakes highlights the need for caution in the development of creative AI systems.

Another concern is the potential for AI to replace human creativity and jobs. With AI capable of generating high-quality content, there is a fear that creative jobs, such as graphic designers, writers, or artists, may become obsolete. This is especially true in industries where cost-cutting is a priority, and AI solutions can reduce labour costs.

The ownership and legal implications of AI-generated content also need to be considered. While DALL-E makes it clear that you own any work you generate, it’s unclear as with many other models as to who owns the rights to AI-generated works, and the question remains regarding whether they can be copyrighted or patented. This creates uncertainty for organizations and individuals who use AI to generate creative works and can lead to disputes over ownership and control.

Creative AI has the potential to revolutionize almost every industry and enhance human creativity. But its use also raises serious ethical and legal concerns. The development of creative AI must be guided by principles that promote transparency, accountability, and fairness across all areas to ensure its benefits are enjoyed by all and its risks are minimized.

When we talk about the applications of these tools being limited only by our imagination, that isn’t an exaggeration. When writing this article, for example, I could have just told ChatGPT to “write an article on the benefits and dangers of generative AI,” and it would have written a decent piece. But to make its writing better and closer to my style and voice, I gave it a few samples first, telling it to read and understand the voice behind them before asking it to write its sections of the article in a similar voice.

As you read the piece, you were probably trying to figure out which parts were AI-generated and which ones were written by me. I could reveal it to you, but where’s the fun in that? The truth of the matter is, if you weren’t 100 per cent convinced that one section was either AI or human-generated, that is enough cause to be excited but wary about the future of AI in the creative world.