Menu

AI Creates: Scared or Sceptic?


Should we be afraid of AI's ability to create? Are illustrators' and writers' days numbered? Sparked by our conversation with a technology thought leader and a magician in issue 9 of Strike, we got thinking, got scared and decided to dig deeper...

By: Clare Howdle,   4 minutes

Sophisticated, deep-learning image generators such as DALL-E, developed by OpenAI, are responsible for the surging wave of AI art flooding the internet right now as brands, publishers and individual punters play with the technology and its capabilities. From Cosmopolitan’s first AI art generated cover in June 2022, to the endless cycle of memes mocking when AI art goes haywire, the software is continuing to causing a stir.

And then there’s large language models like ChatGPT, (also developed by OpenAI), an artificial intelligence-based chatbot that’s been trained to interact with users in a natural, conversational way. From Last of Us showrunner Craig Mazin mucking about with director Rian Johnson to see if ChatGPT could write the synopsis for the sequel to The Glass Onion, to Oxford and Cambridge banning ChatGPT because of its ability to write a degree-level essay, there’s clearly concern about its capabilities and craft when it comes to aping, improving on, or overtaking us as writers.

Both programmes work in the same way. Roughly speaking, they take huge amounts of data, search for patterns in it and become increasingly proficient at generating statistically probable outputs — such as seemingly human-like language and thought, or accurately rendered imitations or styles of artwork (see what happens when you ask it to “create an illustration of a rabbit waterskiing in the style of Frida Kahlo”, for example).

"Right now, it can still be caught out and called out – an image where things not only don't quite line up but look downright unsettling, a sentence that's too vague and repetitive."

Risk or rewarding?

Of course, at present there isn’t an immediate risk to creatives. The AI is not advanced enough to do what highly trained, professional creatives (ah-hem) do, in terms of original idea generation, compelling word craft, or unique authorial illustrative style. Right now, it can still be caught out and called out – an image where things not only don’t quite line up but look downright unsettling, a sentence that’s too vague and repetitive. But the more we use the software the better it will become. Able to replicate different artists or authors, or ape the style or tone of a particular brand or organisation at the drop of a hat. How long before Google’s characterful one-off home page illustrations are AI-generated? Or are they already? Gah.

According to OpenAI, the risk of us all losing our jobs to a machine is still low. DALL-E is a tool that “enhances and extends the creative process”, an OpenAI spokesperson told VentureBeat about the decision to open DALL-E up for commercial use. The articles goes on to elaborate, explaining that “much like an artist would look at different artworks for inspiration, DALL-E can help an artist with coming up with creative concepts”.

However, as Henry Williams puts it in his Guardian piece exploring whether ChatGPT means copywriters are done for, “I don’t claim any superior insight, just a realisation that if a company can improve its bottom line by cutting costs in its supply chain, it will. Any sentimental attachment to human-created content is sure to be quickly overridden, I suspect, by the economic argument. After all, AI is super-fast labour that doesn’t eat, sleep, complain or take holidays.”

He goes on to hit us hard with the numbers. PriceWaterhouseCooper predicts that 3% of jobs are already at risk from AI. By the mid-2030s, this proportion will jump to 30–44% among workers with low education. That’s a lot of people who will need to “upskill”, retrain or drop out of the workforce.

So is it an opportunity for a rethink in how we create and the tools we have at our disposal to do it? Or is it a threat to the way creative businesses work? Time will tell. Only one thing is certain – as always, it is those willing to adapt to change, pivot and innovate that will thrive. Let’s hope we’re all among them…

Read on...

This is the most cursory of toe-dips into the world of AI large language models and image generators, and the spectrum of existential dread or epoch-shifting exaltations that accompanies it. For a deeper dive, check out these very good articles on what it all means and where we’re heading…

Noam Chomsky’s guest essay for the New York Times on the false promise of ChatGPT starts with a Jorge Louis Borges quote before pulling us feet first into the debate about whether the software is cause for concern or optimism. What’s not to love about that?: The False Promise of ChatGPT

Copywriter and journalist Henry Williams takes a frank, personal look at ChatGPT and whether it spells the end for wordsmiths: I’m pretty sure AI is going to take my job

New York magazine photo editor, Megan Paetzhold put DALL-E to the test at how well it could do her job. So… Putting AI competition to the test

 

+

Images generated with the term “Waterskiing rabbit in the style of Frida Kahlo”, using (former) DALL-E and (latter) Starry-AI.

+

From reading our minds to the existence of objective truth, dig into our mind-blowing conversation with a research magician and a technology thought leader in our shiny new print issue of Strike, out now.

Read our original feature about AI taking on humanity in The Magician and The Data Scientist, Strike #1

Also in this issue