I need basics
Ok, I got you. Let’s get you up to speed with just enough info to make you dangerous.
What is it? There are two kinds of AI you need to know about – let’s call them THE ALGORITHM and THE GENERATOR.
The algorithm analyzes patterns and detects relationships in data, then gives you results — like what show to watch next after you binged Floor is Lava.“
The generator is trained on SO MUCH data. It takes what it’s fed, and then when it’s prompted, it goes through everything it’s been taught, looks for patterns that match your input, and uses it to generate a brand new output. This could be language, images, audio, video or even code.
So why is everyone talking about this now? One reason is that the UX for a lot of these generative AI tools got crazy easy to use last year. Like, if you can use a keyboard you can create mindblowing imagery or loads of textual content in moments.
The second is our friends at Open AI who introduced the easier-than-midjourney-on-discord Dall-E (see better UX as described above) and then promptly blew the world’s collective mind with ChatGPT. And in just a few months, the Generative Pre-trained Transformer (GPT) AI was running on everything from search engines to Teams.
Which brings me to why you should care. Because it’s here to stay and it’s just the beginning of how generative AI is going to transform so many aspects of our industry. And you can be the most hardcore straight edge technophobe but at some point — unless you are the absolute hottest of shit, which I sincerely hope you are — you are going to need to learn to live with our new robot friends.
Ok, but how?
For conceptual work, generative AI can be an aid for brainstorming and ideation. Feed it with the brief, tell it a bit about the brand and then play around. It will generate ideas and suggestions that might rule out certain routes -or in the best case - inspire new directions to explore. A lot of the output you get will be things you will just toss out. And you should, there is a lot of rubbish created. One thing I think it’s quite good at is going down certain paths a little ways to see if it’s worth going further - and AI’s feelings won’t get hurt when you don’t go with one of it’s ideas.
Another way I’ve found generative AI can also help is in the actual writing process. It tends to generate terrible headlines and taglines but is better at long-form content, as long as you give it well-defined parameters in your prompt. It needs heavy editing, but it can absolutely save time and even provide new perspectives on how to approach your messaging.
For art directors & designers, generative AI is even more dramatic. Image generators give you a way to visualize an idea without spending time looking for stock or comping together a bunch of images to get your idea across (and yes, even when you get the output you’ll probably have to retouch it to make it work).
But it’s not all image generation from text, there are plenty of AI tools that can help you fine tune your visual concept. Even Photoshop has a tool that lets you change facial expressions, skies or even landscape with just a few clicks. You can also select parts of the image and erase them with one click, leaving the background looking relatively perfect where the thing was. Honestly, if I could have all the time back that I spent cloning out parts of images, or extending them or even making people look “happier!” for client presentations, I would likely have years of my life back. I envy the junior ADs of today (shakes fist in old).
Always remember to use generative AI as a starting point, or a sketch and build your own version from what it’s given you when it comes to finalization. Unless your idea is to have generative AI play some sort of role in your concept and you want to feature it as-is, of course.
I’m going to go over a few of these tools in detail, so stay tuned for more.
