4 ways generative AI will change marketing (with links to tools)

In the wake of Generative AI's viral year, Large Language models are starting to have a moment, as ChatGPT is officially the fastest growing product of all time -- and with the release of plugins, is giving humans superpowers.

And last week the AI hype train hit a fever pitch, evidenced by a San Francisco meetup that many are calling the "Woodstock of AI", which was held by Hugging Face and required several venue changes to meet demand.

The power of AI is starting to become clear and mainstream, and many have their eyes intently on the future.

With regard to paid ads, I see the future as a closed loop system, where AI analyzes ads, then strategizes, produces, and optimizes the results, before repeating over and over, with only minimal human assistance – in the same way that derivatives trading consolidated from thousands of traders to a handful of firms after going electronic in the early 2000s. We already see some initial companies such as Creative.ai moving in that direction, and I could imagine just a handful of creative producers owning the lions share of impressions on Facebook.

But that's the future – and at the moment, it's not immediately clear how advertisers can capture value.

As a result, over the last few weeks I've been working on a few MVPs to find novel use cases for AI in ads. The results have been encouraging, with even weird or robotic concepts looking promising – and just this week a client of mine achieved a nice win with a concept that stitched together several different AI tools.

Now, after a fair bit of research and some proven tinkering, I am sharing four predictions about how generative ai will impact marketing, plus tools that are already having an impact – to help you get started.

1. UGC/Influencers will be disrupted quickly, especially for still photography.

It will become easy to generate, recreate, and control the environment and poses of photorealistic "influencers" or other traditionally human models, either with out-of-the-box prompting or with a combination of AI tools and traditional software, such as 3D modeling.

Part of the value of UGC/Influencer ads is leveraging appearance, which will erode quickly. With the latest releases of MidJourney v5 and stable diffusion, creating photo-realistic images is easy, and most of the obvious signs of AI have been smoothed out. Tools such as ControlNet allow for consistency and manipulation, allowing advertisers to create and control digital personalities with ease.

2. Brand Marketing and Product Photography
Physical shoots and models are expensive and bounded. As above, it will be easy to build basic 3D models, set the scene, and then manipulate photo-realistic product photos from reference images.

3. Voice Overs
Tools such as Eleven Labs allow for training of speech with just one minute of audio. As a result, it's possible to shoot with an influencer or voice talent once and then iterate using their voice perpetually. This will significantly erode the market for voice talent.

4. Video and Animation
Basic character animations from still shots are already available (d-id.com), although still pretty robotic.

More sophisticated advances in AI animation and video exist by way of stitched-together collab or Hugging Face notebooks, although still pretty raw. I predict that more advances in animation and video will come down the pipeline shortly. I imagine a combination of reference video + GAN models or 3D + GAN for manipulation (similar to still shots from prediction #1).

For a good AI in marketing follow, check out Manson Chen's newsletter, Parabolica

If you are interested in learning more about AI in advertising, book a 15 minute call for a creative consultation.