AI Made Making Things Easier. Having Taste Got Harder.
AI has made it incredibly easy to make things. Writing, images, videos, websites, presentations, ad concepts, social posts. Give the right tool a prompt and within seconds you can have something that would have taken considerably more time a few years ago.
I'm not convinced that's entirely a good thing.
I use AI regularly, but I've developed a fairly clear line around where I think it's useful. I like AI as a tool for working on something. I don't particularly like it as a tool for making the thing. Using AI to brainstorm twenty headlines so I can find an interesting direction? Useful. Using it to help restructure something I've already written? Great. Removing an unwanted object from one of my photos in Adobe, extending a background, changing a crop or helping with repetitive production work? I'm completely on board.
Typing "make me a cool brand" and taking whatever comes out the other side? Not so much.
Beyond the obvious questions around originality and creative ownership, I don't actually think that's a particularly efficient way to work. AI can generate almost unlimited options. The problem is that most of them are fine.
And fine is becoming incredibly cheap.
The First Draft Is Basically Free Now
This might be the biggest change AI has made to creative work. The first draft used to have a cost. You had to sit down and write something, sketch something, photograph something or open Photoshop or Illustrator and start moving things around. You had to make decisions before there was anything to look at.
Now you can skip that uncomfortable part and immediately generate something competent. That's appealing, but it's also where I think things start going wrong.
When the first draft comes from you, even if it's terrible, you've already started making decisions. There's an idea underneath it and some intention behind it. You're solving a problem rather than selecting from a menu of solutions somebody else handed you.
When AI generates the whole thing, your role starts shifting from creator to curator. Curation is useful, but it's not the same thing. I'd rather use AI after I've started than ask it to start for me.
AI Is Great at Brainstorming
This is probably where I use it most. Sometimes you know there's an idea somewhere, but you haven't figured out exactly what it is yet. AI is very good at throwing possibilities at the wall.
I'll use it for things like:
Give me twenty headline directions.
What am I overlooking here?
What's the counterargument?
What's another way to organize this?
What questions would a customer have?
Where is this campaign idea weak?
Most of the answers won't survive. That's fine.
The value isn't always the answer itself. Sometimes it's seeing something slightly wrong and realizing why it's wrong. That sends you somewhere better. It's basically a brainstorming partner that never gets tired and has absolutely no emotional attachment to its ideas.
Very convenient.
Copywriting: Help Me Edit, Don't Write It for Me
AI-generated copy has improved enormously. It can produce perfectly competent marketing copy in seconds. That's also exactly why so much writing online is starting to feel the same.
You start recognizing the rhythms. The neat little contrasts. The endless headings. The dramatic one-sentence paragraphs. The insistence that everything is a "game-changer." Apparently we're also unlocking quite a lot of things.
For writing, I find AI much more useful when there's already something to work with. I can write a rough paragraph and ask for help making it tighter. I can use it to find repetition, explore alternate headlines, reorganize ideas or point out where an argument isn't clear.
But I still want the original thought to be mine. Otherwise, what am I publishing?
If the reader could type the title of my article into an AI tool and receive essentially the same article, I haven't contributed much. This matters even more as AI content floods search results and social platforms. Having an actual opinion, experience or perspective becomes more valuable when generic information is almost free.
Image Generation Is Where I'm Much Less Interested
I've experimented enough with generative imagery to understand why people like it. It's fast, sometimes it's genuinely impressive, and every few months it gets noticeably better. But generating a finished image from scratch isn't something I'm particularly interested in making a core part of my design process.
There are practical reasons for that. You can spend an absurd amount of time trying to make a generated image look like something that could have simply been photographed or designed properly in the first place. Fix the lighting. Fix the hands. Fix the reflections. Make the architecture less perfect. Make the imperfections less perfect. Remove the weird object in the background. Fix the thing you didn't notice until somebody else pointed it out. At a certain point, "AI is faster" becomes questionable.
There's also something creatively unsatisfying about endlessly prompting a machine until it produces something you like. You're directing, certainly, but you're giving up a lot of the decisions that normally happen while making the work.
That's not where I think AI is most interesting for designers.
Adobe's Version of AI Makes Much More Sense to Me
Where AI becomes genuinely useful is when it's integrated into tools I'm already using. Adobe has increasingly built generative features into creative workflows, and this is much closer to how I want to use the technology.
I've taken the photo. I've made the design. I've chosen the composition. Now there's a garbage can in the background that I don't want. Remove it. The image needs to become portrait instead of landscape and there isn't enough background. Extend it. A product shot is right, but there's something distracting in the corner. Fix it. The building would look more striking in blue. Paint it.
That's a fundamentally different relationship with AI. I'm still making the work. AI is helping me manipulate it.
Generative Fill, Generative Expand, object removal and similar tools can eliminate genuinely tedious production work. Things that once required careful cloning, masking and rebuilding can sometimes happen in seconds. That's useful automation. It's not asking AI to be the photographer, art director and designer simultaneously.
The Same Goes for Marketing
There are plenty of boring parts of marketing that I have absolutely no sentimental attachment to doing manually.
Summarizing campaign performance? Go ahead. Give me variations of an email subject line I've already written. Pull themes out of customer feedback. Help organize keyword research. Turn a messy page of campaign notes into a usable outline. Identify questions I'm not answering on a landing page.
Those are good uses of AI because they give me more time for the parts where judgment actually matters.
I don't need AI to invent a marketing strategy and hand it back to me in six bullet points. Strategy requires context. What has worked before? What hasn't? What's happening inside the business? What does the customer actually care about? What can we realistically execute? What does the brand sound like? What does everyone involved know that isn't written down anywhere?
AI can help interrogate those questions. I don't think it should automatically answer them.
Taste Is Becoming More Valuable
This is the part of AI that I find genuinely interesting. When almost anyone can produce something technically competent, technical competence stops being much of an advantage. You can generate a polished image, grammatically perfect copy, a clean website or fifty logo concepts before lunch. So what?
The difficult part becomes knowing which image feels wrong. Which sentence sounds generic. Which logo doesn't fit. Which idea is derivative. Which campaign shouldn't exist at all.
That's taste.
And taste isn't some mystical creative gift. It's something you develop from years of making things, looking at good work, making bad work, receiving feedback, noticing details and figuring out why something works. AI can imitate the result of those decisions. I'm much less convinced it can replace the experience that teaches you how to make them.
More Isn't Better
One of the strangest promises around AI is productivity. Now you can create ten times as much content!
I'm not sure anyone asked for that.
We already have more content than any person could possibly consume. Brands don't need to publish twelve mediocre posts instead of three good ones. Designers don't need 200 concepts when five thoughtful ones would have been enough. The internet isn't suffering from a shortage of things.
If anything, AI makes restraint more important. The ability to generate unlimited options means somebody still has to say no to almost all of them.
Where I Draw the Line
My rule isn't particularly complicated: AI should help me do my work. I don't want it replacing the part where I actually make the work.
I'm happy using it to:
Brainstorm and challenge ideas
Research and organize information
Edit and tighten my own writing
Analyze marketing performance
Explore variations of an existing concept
Resize, extend, modify or clean up my own imagery
Remove repetitive production work
I'm considerably less interested in asking it to generate the finished idea, article, photograph, design or campaign and then putting my name on it.
I'm also not pretending there's some perfectly clean boundary here. These tools are changing quickly, they're being integrated into practically everything, and we're all figuring out where they fit while using them. My line will probably move. But I don't think the future of creative work is simply generating more stuff, faster. We've already solved that problem.
The more AI makes competent work abundant, the more valuable the things it can't conveniently generate become: perspective, restraint, context, experience and knowing when something is actually good.
Making things is getting easier. Knowing what deserves to be made might be the harder skill now.
