Generative AI is for going deep, not wide

Last month, I shipped out to Las Vegas to attend this year’s Adobe Summit. A few of my colleagues handled more overarching recaps of the event. As for me, I’m stuck on one particular theme that clearly stands out. Naturally, that’s generative AI.

Groan. Eye roll. Stick with me.

Let’s start with this: Adobe, just like Microsoft, Google, and other tech giants, appears to be making great strides toward integrating generative AI capabilities directly in to its tools. This is kind of a duh revelation, but, in my estimation at least, it is the easier-said-than-done development needed to mainstream AI for, well, for people like myself.

Put another way, at this exact juncture, generative AI has almost no bearing on my day-to-day professional life. There are a few reasons for that, but the biggest is quite practical. To use ChatGPT or Midjourney, I have to leave Adobe Experience Manager, or Figma, or Microsoft Word, or whatever application I’m actually using at that moment, put in my prompt and cross my fingers. If all of that goes well, I have to save the image or copy and paste the text, and by the time I’ve round-tripped back to the tool I was actually using when all of this started, well, I haven’t really saved myself much, if any, time. And that’s not even broaching the topic of ethics or legality or brand safety as we weigh up generative AI’s utility in its current form.

Anyway, Adobe seems deadset on changing this calculus. Almost every session I attended featured some showcase of these tools going native in a way that goes well beyond Firefly finding its way in to a few more Creative Cloud applications. If you, like me, are tasked with managing brand consistency while also enabling dizzyingly voluminous scale, then Adobe’s introduction of Custom Models, which will allow Adobe’s AI to be trained on your brand, is the biggest deal of all. Not only will AI be native to Adobe’s tools, but it will also know me and my brand.

Not coincidentally, this is where this starts to get exciting on the day-to-day level. And that is because the real power of generative AI, as showcased over and over out in Vegas, is in the endless remixing of what you already have on hand - your logo and color palette and visual style and owned assets and so on.

Do the math. You don’t actually need very many raw materials to get to hundreds of thousands or even millions of combinations. In fact, you probably have the basic ingredients to get there right now. What you don’t have - at least not yet - is a machine to help you rapidly assemble and re-assemble all million of those combinations in a matter of minutes.

We humans are constrained by the laws of time and space. Not so for generative AI. It does some very strange things when given an almost blank canvas - its six-fingered men and casually told lies belying its lack of humanity. But when given some guardrails, its potential to do what those of us who are corporeally challenged can not seems likely to be cause for celebration. And, wonderfully, that capability does seem to be at our fingertips now.

Sure, generative AI still has a role to play in other parts of the creative process. It can certainly serve as a built-in riffing partner when you are ideating. But, at least to me, it’ll be a much more exciting though slightly less magical development when it can instead be handed a fully-formed idea and give it legs that stretch from social to email to website to out-of-home advertising and beyond.

In this context, it seems almost silly to worry about the machines taking jobs. What they’re really coming for is a job none of us has the time or spirit to do. We can go wide - maybe wider than ever before - coming up with a winning idea, safe in the knowledge that once we’ve stumbled upon just such a thing, the machines will be able to help us go deep.