When you think of AI and GPT-3, “creativity” is probably not the first word that comes to mind.
But that’s not to say that the output of these tools isn’t impressive.
In fact, the quality of AI-driven content can be quite startling.
Sometimes (actually, very often) it’s indistinguishable from human-generated content.
Whether or not that process is “creative” as humans have understood it for thousands of years is another question.
After all, these systems are only regurgitating a facsimile of what they’ve already been trained on — aren’t they?
But where does creativity itself come into play with AI, if it comes into play at all?
Let’s look at a few ways AI is already operating in traditional arts sectors, and what it might mean for the future of those creative disciplines — including storytelling and content creation.
Artificial Intelligence in the Creative World
It may seem counterintuitive to talk about the place of AI in fields centered on human creativity.
But the fact is, it’s already here.
In 2016, 20th Century Fox partnered with IBM to develop an AI-created movie trailer for its horror film Morgan.
Researchers fed IBM’s Watson platform hundreds of existing horror film trailers, and then had it select scenes from the completed movie for the editors to splice together.
In 2018, an AI-generated piece of art, Portrait of Edmond de Belamy, sold at a Christie’s auction for over $400K.
The AI algorithm that created the artwork was trained on a data set of 15,000 portraits that spanned six centuries.
More recently, software engineer Phillip Wang spun up a website to show how AI can be used to create extremely realistic portraits of people who don’t exist.
(Wang has created a similar website for cat photos, if that’s more your thing.)
Personally, I couldn’t tell the difference.
The same trend is happening in music.
Researchers at Sony used the company’s Flow Machines software to create the first pop song written by AI, based on a dataset of 13,000 lead sheets from different musical genres around the world.
AI is, already, firmly embedded in the visual and musical arts.
So it’s no surprise to see it operating in the storytelling world too.
In 2019, filmmaker Kevin Macdonald (of The Last King of Scotland fame) directed a Lexus commercial using an AI-created script that had the car flying past panoramic scenery before saving itself from a fiery crash
AI and Robots Can Replicate — But Can They Create?
The common denominator in all of these examples are machines that are fed a wide array of sample data (including art)…
And then using its algorithms to generate output that looks and sounds a lot like a human created it — often with remarkable results.
But are these AI systems engaging in what we might call “creativity”?
Whether the output is good or bad, these systems are producing something that hasn’t existed before.
AI models new content based on patterns and themes in existing works, which isn’t so different from the way that humans engage in creativity.
After all, even the most innovative artistic creators usually have a solid grounding in the tenets, theory and predecessors of their chosen form, whether it be classical piano or creative fiction writing.
The devil, as always, is in the details.
While a core part of creativity is taking familiar elements and combining them in new and unfamiliar ways, it’s usually informed by a guiding theme or principle.
There’s meaning and intent behind it, at least to the artist.
The challenge with AI systems is that their goal is to produce something that reasonably matches what they’ve already seen.
They may produce something novel, but that’s more likely to be due to the inherent randomness of their deep-learning algorithms than any sophisticated “thinking” or inspiration.
One academic found that the written word seems to be the hardest medium for AI to demonstrate real creativity in.
For one thing, the meaning of a sentence hinges on more than just the meaning of the words.
A sentence takes on different meaning depending on the context of that sentence as well as the background of the reader.
Other experiments with GPT-3 have shown that the longer the text, the more incoherence creeps in.
AI algorithms can produce coherent text in shorter chunks, but doesn’t do as well in maintaining a larger narrative arc.
So much of AI’s success comes down to what you’re asking it to do.
Artificial Intelligence Can Augment Creativity
I think all of this is good news for writers and other creators.
A big part of human creativity involves exploration and experimentation.
And this requires absorbing a lot of information, ideas, and models — and then putting it all together in new ways.
This is where AI has an advantage.
AI systems can digest exponentially more data than any human ever could. They have more information and more models to play with.
So whatever the creative pursuit — visual arts, music, poetry, fiction writing, movie plots — AI can do some serious heavy-lifting to surface new combinations of existing forms and language, in ways that one human cannot.
That doesn’t mean AI will automatically produce a new, perfectly formed work of art (or story or piece of music), ready for human consumption and appreciation.
But it does put us way ahead in terms of experimentation with existing language and formulas.
The possibilities here are endless, not least because these AI platforms don’t always generate the same output from the same input.
Case in point: One writer used the same three prompts in GPT-3 for “ideas to try out with GPT-3” (very meta) and got a different list every time.
If you have time to sift through a lot of AI-generated output, you just might see something that sparks your own creativity.
And using that as a jumping off point is no different from being inspired by a revered (or just eccentric) predecessor in your field to generate something new for your audience.
In any creative endeavour, think of AI as you “able assistant,” not as your rival.
AI and the Future of Your Creativity
There’s no question that AI can help jump-start a lot of creative projects.
And that includes your content marketing.
AI platforms may eventually become our collaborators.
Machines can’t emulate emotions (the core of creative work), but they can help accelerate the experimentation and prototyping we use when we’re looking for new ways to reach people.
AI can put random concepts together, but ultimately, humans still need to create the stories that connect them.
Humans still need to provide the meaning.