Copyright, Differentiation and Creativity in the Age of AI

I had a question for a lawyer the other week as we, like many others, ponder the immediate implication and application of the latest AI tools that are coming to market. My question related to copyright in the age of AI is this: are there copyright implications to using content generated from AI?

This is significant should we consider using AI in a client context. My line of thinking was that if the AI was trained on human-generated source material, could it be argued that derivatives of that, in the form of output from the AI, constitute infringement?

The answer was as simple as it was surprising.

Only human output / product can be copyrighted. So anything a machine generates is impossible to copyright (under the current laws and understanding). There are grey areas here such as AI applications that have been trained on known and recognizable works from famous artists. The initial outcry from these creators is that AI is emulating their style by consuming the images of their art and then using that material to generate new imagery in the same style.

Interestingly, Adobe’s AI engine, Firefly, was trained on images that Adobe already has licensing agreements for. So anything you generate out of AI versions of Photoshop is clear from a rights perspective, as Adobe has provided compensation to the creators, for the source materials.
So copyright can’t be granted to (text-based) machine outputs. And if we have visual assets they are okay, especially if generated from a model/platform that is paying some kind of royalties or licensing fees. Noted.

Passing The Differentiation Test

A few weeks back Google announced that it would release AI-enhanced features into its ad platforms, including the ability to serve up AI-generated ads. It was reported that media managers would be able to choose which AI-generated ads they wanted in the campaign, providing a level of oversight. One assumes that the ad platform is inferring appropriate imagery and copy based on the category that company operates in, along with some basic understanding of the product offering.

Since the rise of Machine Learning (ML) a decade ago, my business partner and I have predicted that much of the actual work our firm puts into building and running campaigns for clients would be displaced by the machines. It’s a natural extension of a function that’s primarily analytical. The way we normally operate is that we establish campaigns with a variety of ad formats, creative and placements, and then analyze what ads and placements are working or performing best for our client’s goals. We then make adjustments to optimize towards that end. Machines are great optimizers, especially when empirical facts are involved.

But this new AI-enabled feature of Google Ads (only being rolled out in the US for now) takes this one step further and begs the question: if a big tech, ubiquitous ad platform like Google was presenting art and copy suggestions based on a basic understanding of the business, how would the ads generated for one brand in a category be distinct or different than what the ad platform serves the competition?

This problem isn’t limited to Google’s Ad platform, but rather highlights (I think) an inherent risk in using generative tools in client work and contexts. If you’re in the business of differentiation (as anyone who works in branding and marketing field is), using generative AI is (currently) most likely to give you an average output, not a particularly differentiated one.

Materializing the Value

AI is new tech and, as I’ve already written, the general hype of its possibilities are getting much further ahead than the actual development of the technology itself. While we collectively expect AI to improve and evolve, its current value to the market can only be manifested if we’re applying it in contexts that make sense. Remember— it’s still just a tool, and a tool needs an adept human operator in order to materialize its value.

So where does AI add value? Clearly not on anything that needs to be differentiated, or protected (from a copyright perspective).

Have you heard the story of Thomas Edison and his quest to create the incandescent lightbulb? Legend has it that he went through over three thousand prototypes before he finally found the right mix of materials. When asked about all of his failures, he famously replied that he didn’t fail–he just found a few thousand ways that didn’t work.

I don’t know how true this story is, and I’m too lazy to fact check it, but this anecdote comes to mind when I think about how I’m using AI right now from a writing perspective. In general, I find it pretty anemic in its ability to generate something worthy of being read. But I do think there’s some value in helping me figure out what I don’t want, what doesn’t work.

For a recent piece, I asked it to generate sub-headings for the sections of the article. One part of the piece was about timing and it fed me bad cliches about “time” in general. Now I write a lot, and in a professional context, I write brand strategies—turning a phrase comes fairly naturally to me. It’s not that I have a problem coming up with headings, but I was tired, and the act of penning something as functional as a subheading seemed mundane, so I decided to let AI have a go.

It failed miserably at getting me what I needed. But what it helped me do was figure out the avenues not worth taking, to let me arrive at the right one. In this respect, perhaps, it helped me speed up my own creative process–if only ever so slightly.

Creativity and AI in Practice

The creative process is not linear or formulaic. Yes, some creatives have their process, but I’ve never met a creative who could distill that process down to a level of mechanical consistency. But that’s what AI still is—mechanical. The writing sounds mechanical. The visual outputs, while novel, are effectively mechanical variants of what’s already in existence.

What AI can give us, right now, as creative professionals, is ways to speed past the dead ends and cul-de-sacs of the creative process. And let’s acknowledge that there are ALWAYS dead ends in the pursuit of the right creative solution.

The right-sized application of AI is not designed to replace human creativity, but rather to use it as a tool to augment. It shows us the mechanical permutations of something that would otherwise be time-consuming for us to explore.

It’s my sincere hope that we don’t start to see a world of brand and marketing that’s a sea of sameness, generated from the AI of the most popular models. But instead, we see a tidal wave of unique and interesting work that was achieved with far less blood, sweat and tears than it would have otherwise taken using traditional tools.

Spencer Saunders

President & CEO