Third of a Three-Part Series
This is the third of a three-part series on AI-driven content creation.
- Part One: The Debate
- Part Two: Perspectives from Creative Pros
- Part Three: The Path Forward
Part Three: The Path Forward
In Part One of this article, I positioned the Debate. In Part Two, we heard from a wide range of talented creators. For this third and final installment, I’ll share my thoughts about the path forward, including how we can integrate AI assistance into our creative workflows.
Toolsmiths, if Nothing Else
Since the beginning of humankind, the ability to invent and use tools has separated our species from all others. Whether it was a sharpened flint used for scraping and cutting animal hides, wheels and axles to carry loads, a magnetic compass to navigate the globe or the printing press to communicate ideas to the masses, tools have played a crucial role in our efforts to elevate the human experience.
Sometimes, a new type of tool offers such radical improvements that it quickly renders entire professions obsolete. When was the last time you crossed paths with a blacksmith? Have you ever hired a professional lamplighter, stone miller, trapper, cobbler or candlestick maker? Me neither. And if you do know any, they’re likely doing business based on the sheer novelty of keeping some quaint but dying craft alive.
But AI raises new, more profound questions. A big one: is it inherently dangerous? For the first time in history, serious people are having real debates about the potential of an emerging technology to take over and run amuck. And the perceived dangers go way beyond old-school accidents caused by some unfortunate human getting sucked into some newfangled piece of machinery.
Personally, I’m not losing sleep over the whole “robots replacing us” hysteria. AI content assistance is a toolset created by humans for humans. We can use it to help find creative solutions faster and more accurately. But we’re still in charge.
I’m still thinking back on comments from Part Two of this series. They provide useful perspectives for those of us trying to make sense of the brave new world ahead.
“Obviously, the answer to the question is BOTH (tool and threat)...”
“The genie won’t be going back in the bottle, so there’s no turning back...”
“AI is evolving quickly and will impact our jobs and careers…”
“...we need to pay attention to what is happening now to evolve and adapt.”
“AI is not going to take your job. A person using AI is…"
“…art is communication to self; design is communication to others.”
“I think AI could be useful to artists who embrace it as a tool…”
“Artists are inspired and driven by a passion to create. An algorithm isn’t.”
“If there’s one thing that impresses me, it’s how quickly this whole field is evolving. The way users approach it will continue to evolve as well."
“For years it’s been only theoretical; now it’s real.”
The Human Factor
While AI may be great at certain tasks, there are specific (and important) design roles that still require the human touch.
Inspiration and Instigation
Sure, AI tools can spit out countless iterations in short order — when prompted to do so. But curiosity and imagination are uniquely human traits. AI will never have the innate ability to create something that’s never existed before. The creative spark is intangible and mystical — God's gift to humanity.
The David Carson Experiment'
In Part Two of the series, Nick Ugre shared, “Yesterday, I saw a poster that was created by AI in the style of David Carson, and it was better than many junior designers' work.”
I mentioned this to David in a chat. He expressed interest in seeing the images. I relayed this to Nick, who couldn’t remember exactly where he saw them. So I challenged Nick to recreate what he saw. The first set of three images above were created by Nick in DALL-E. The second set of three images are actual David Carson graphics.
No disrespect to David, but it seems to me, anyone familiar with his graphic style would likely believe all of these pieces were done by him, if told so. But here’s my big point: to achieve these results, there first had to be a David Carson. A David Carson style had to already exist, and DALL-E needed someone to supply the specific prompt.
David Carson Comment
David’s own take? “Ultimately, it’s about the designer’s eye and intuition...so the same people who use clip art, stock photos, etc. (the ones with no innate design sense) will find some use for this. And mid-level agency designers should be a little bit concerned. : o”
AI content is only validated by human discernment. Without human validation, AI content lacks any value at all. Nor does AI possess any sense of propriety. Human curation — thumbs up vs. thumbs down — will always be essential.
My previous comment in Part Two about the objectification of women in AI-generated content is quickly, and sadly, born out in this next Sidebar.
(Sexism in AI: an Experiment)
Note that the word “women” isn't in the prompt, let alone wardrobe and camera angle. The point I’m making is not one of morality, but rather the inherent bias portrayed in these images.
I wondered whether “steampunk” is just a naturally charged descriptor – a term bound to conjure salacious imagery. To put this theory to the test, I did a Google image search using the exact same prompt. While the results still feature women modeling steampunk-inspired fashions, the results are considerably more modest.
Artists have a point of view. Even thousands of years after it was created, a cave painting still conveys the artist’s personal story and point of view. AI algorithms don’t have a point of view, and they certainly have no intention. AI can’t be the author in a creative process any more than a pneumatic nail gun can take on the role of a carpenter in building a house.
Collaboration and Gestation
When I prompted ChatGPT to write the closing in Part Two of this series, it did the job in about 30 seconds. But there was no feedback loop – though I could rewrite the section however I pleased, I’d be doing my editing in a vacuum. And ChatGPT wouldn’t care either way.
Contrast that with my human copywriter colleague, who got back to me with a draft a day or so later. I found his version superior from the get-go. And when I suggested a couple of edits in Google docs, I received a quintessentially human response: my phone rang. The writer wanted to discuss the changes I’d made. After five minutes of back and forth, we had a solution that was better than what either of us had written independently. We all have deadlines, but when the schedule allows for it, gestation and genuine collaboration yield a better final product.
Practical Tools and Professional Workflows
But enough about AI making art and replacing humans. What excites me is exploring how we can leverage AI to amplify and improve our creative output. Much like an extra-skeleton bodysuit helping a paraplegic walk again, AI may be most useful in helping creative types get more done in a day, all while allowing more iteration and experimentation than ever.
The specific task might be upscaling an image to achieve better resolution. Or removing a blink and adding a smile to an otherwise amazing photo. Or tweaking the lighting in a film scene from daytime to moonlight. How about auto-assembling a first cut from your movie dailies? Proofreading and copyediting a novel? Crafting a movie pitch? These AI capabilities already exist.
We’ve mentioned many AI tools in this series, including, ChatGPT, and two of my go-to's: Grammarly and Topaz Image tools. But a new breed of AI platforms is just arriving on the scene. These new products are more like AI workshops than tools with a full suite of task-specific apps. You.com and Runway are two examples.
You.com offers several functional modules: You Write, for copywriting; You Imagine, for AI-generated imagery; and AI Search, touting it as “The AI Search Engine You Control” — a not-too-subtle jab at Google.
Runway offers dozens of tools for audio, video, images, and text. There are specific applets for background removal, text to texture, erase and replace, upscaling images, creating variations on existing images, video transcription, generating movie subtitles, colorizing, motion tracking, adding a greenscreen background to scenes shot without greenscreen, cleaning audio and replacing silence, just to name a few. Runway also provides motion graphics and video templates, plus galleries of filters and effects.
Some of their more exciting tools involve AI training, where you can choose the images used to inform the AI algorithms.
The Bottom Line
The tools will continue to evolve. We must evolve with them or risk being left behind – the choice is ours. If AI is to become a tool in your professional arsenal, you’ll need to invest the time to master it, just like any other tool. And by mastering it, I mean obtaining predictable results. Simply generating countless iterations and hoping for happy accidents is no longer enough. You’ve probably heard that it takes 10,000 hours to master any skill. There’s no reason to think AI assistance is any different.
Perhaps we can tear a page from the Rennaissance masters’ playbook...
Original Comps at the Speed of AI
I’m particularly keen on the idea of custom AI training. It seems reasonable that any artist with a mature style can train the AI using their own original imagery. This would allow AI to assist the artist to quickly comp up variations on a theme before deciding a final direction to pursue.
The great masters often engaged apprentices to assist with larger projects. Michelangelo actually contracted a foreman and four assistants to help paint the ceiling of the Sistine Chapel 1. Would any of us question whether he was the author of that masterpiece? Not likely. If he were alive today, what might Michelangelo himself think about AI assistance?
What Do Customers Think?
Up to this point, we’ve only framed this debate from the perspective of creative practitioners. And for good reason – I’m not sure there’s much debate or concern among our customers. From their point of view, this new way seems better, faster and potentially cheaper. Hey, what’s not to like? But there is one huge caveat: we all need to keep an eye on the legal challenges I mentioned to make sure we aren’t exposing our clients (or ourselves) to lawsuits over potential copyright violations. For the faint of heart, this issue alone may be reason enough to pass on AI assistance for now.
Where do we go from here?
While this three-part series may raise as many questions as it answers, hopefully, it has gotten you thinking. No matter what your take on AI-assisted content, either as a creator or consumer, this mind-bending technology is definitely here to stay. And, like it or not, you’ll encounter more of it in the days ahead.
It seemed only fitting to give ChatGPT the job of wrapping this piece up. Just for fun, I asked it to create a bit of AI-themed Haiku poetry. You can judge the results for yourself while pondering the meaning of creativity as we know it. Personally, I did find it amusing that the robot doesn’t seem overly concerned about maintaining the 5-7-5 syllable count that defines Haiku.
In the meantime, thanks for following along as we explore the fascinating future of AI. This story is far from over.
Humans create stories
AI content stirs emotion
Machine learning at work
Generating content ideas
Suggestions of AI content
Inspiring ideas emerge
I want to express my thanks to the many, many contributors, both on-record and off, who freely shared their insights and knowledge to help make this series possible. And to The Roster content team for their incredible (and entirely human) effort. Thank you.