If you use AI for your product video, I respect your company less

Imagine if the first time you spoke to a potential vendor for your company, you were directed to leave a message on their answering machine. Or if you went to a fast food restaurant that didn’t have a single picture of their food options – only drawings. Would you respect that brand? Would you eat that burger?

I’m going to hazard a guess and say you’d look elsewhere. But even now, AI developers are busy cooking up a wheeze along these exact lines: AI product testimonials.

Using a generative AI video model, it’s now possible to generate short clips of an almost photorealistic video presenter of the kind you might find in a corporate training package, product demo, or video tutorial.

Some of the leading companies in this space boast that their tools allow users to turn a video script into an entire video. Some have taken this process a step further, encouraging users to generate said script using a large language model – a complex, expensive, environmentally damaging journey from a set of bullet points to a full product video.

I’ll cut to the chase. This is a cheap, thoughtless approach to marketing and in time could come to be treated with utter derision by customers.

Simply put, if you haven’t bothered to front one of your customer or communications team in your product video – or even to pay an actor to deliver lines convincingly – why should I trust you haven’t cut corners with your product, too?

Technological limitations are irrelevant

Ah, but Rory, you say. In time, all of these issues from the glassy eyes and waxy skin to the animatronic-like mouth movements will be fixed by more advanced video models. To which I say: so what?

The issue is not with the sophistication of the false product but with the very fact that it is false.

If issues such as the uncanny valley can be overcome – and there is no evidence to suggest they can – all that leaves us with is a scenario in which no one’s customer content can be trusted.

Let’s not start down this path, which will see customer trust gradually erode to the point that all product imagery, videos, and even tutorials will be thrown into doubt.

The same is true for AI product images, one of the earliest concrete business use cases for AI image generation. I’ve been to numerous keynotes that saw executives cheerfully show me an AI-generated image of some consumer products – from clothing to camping supplies – and praised it as the future of the marketing department.

I find this profoundly disturbing.

Let’s be clear: this is not the same thing as using CGI product images in marketing, a common practice from the smallest businesses to high-end brands. Said images are generally taken directly from the computer-assisted design (CAD) files used in the production of the items themselves, rather than an AI approximation of

Instead, this is generating entirely fake images of products that may or may not exist. There’s both an accuracy gap – any AI product image is likely to have small imperfections that mislead consumers – and a significant trust gap here.

What are we doing here, exactly? Hire marketers, hire photographers, hire actors (or platform your in-house experts) to sell your products and services. Firms that continue down this path don’t risk loss of credibility, but all respect.