Photography in time of AI

Does Photography Still Have a Place?

We've all heard the claim: photography is dead, generative artificial intelligence rules and is all you need. While there's some truth to this statement, there's far more nuance beneath the surface.

Admittedly, in areas well-covered by training data, implementation will move faster—for example, generating B2C graphics for fashion social media. And while it's easier, a well-prepared, repeatable fashion content campaign created with genAI still requires significant effort and business input. It also won't work flawlessly 100% of the time and still requires human oversight.

When you enter the B2B space, especially industrial equipment manufacturing, things quickly get complicated. Popular models have far fewer comprehensive datasets covering equipment and machinery. The reason is straightforward—there simply isn't enough data to train models on. Moreover, the limited data that AI companies managed to gather from the internet often came from very generic stock photos or low-quality images posted on manufacturers' websites.

AI or Photography? It's Not a Simple Question.

The issue is further complicated by the fact that data doesn't work without training context. Simply providing photos isn't enough. A generative model must "understand" (that's a significant simplification on my part) what it receives in order to "transform it back" into an image. While this is fairly easy and automated for rich datasets of people or animals—machine learning can fairly easily identify what a "hand," "ear," or "eye" is (though everyone likely remembers how complicated it was for genAI to understand what human hands should look like)—the same cannot be said for industrial equipment.

Now imagine what training a model for your niche food processing solution must look like. Yes, that's exactly why when you tried generating images for your campaign, the model never hit the mark. If at all.

Of course, there are many techniques available, but ultimately they all require real input data. And unless you're using your own local model—do you really want to hand over your company's confidential information to an AI model? Yes, most AI companies claim your data is safe and won't be used for training. However, remember that these same people simply scraped copyrighted data from the entire internet without paying a dime for it.

Ultimately—for an industrial equipment manufacturer operating in the B2B sector, a quality photography session and product visualization remain important tools for communication and narrating your company, quality, and solutions.

Are We Returning to the "Real" Approach?

There's already research suggesting that people value AI-generated advertisements less compared to human-created content. Gartner Inc. reports that 50% of American respondents opposed partnering with companies using generative AI in advertising because they question the "realism" of the advertised products.

Research from the Nuremberg Institute for Market Decisions also indicates a clear tendency toward skepticism regarding companies using AI in brand communication and advertising.

One could argue these studies address the topic from a B2C perspective, but do we really believe that a person unwilling to trust and purchase a product advertised using GenAI will trust a B2B brand presenting its products the same way? That it won't raise doubt and distrust?

Using Kahneman's thinking model, we could theorize that a rational business client should remain on the path of slow, analytical thinking, thereby checking all available data and understanding that AI-generated content serves merely as a "hook" designed to catch attention or simply as attractive visual filler.

However, a business client is not some otherworldly entity—they're represented by people. And that means fast, heuristic-based thinking, at least in the initial stages of client interaction. This leads to a simple conclusion—if heuristic mechanisms suggest to the client's representative that AI-generated content should be treated with suspicion and may indicate lower quality… we could be looking at a very poor start.

Mixed Reality

We must, however, look at this realistically: companies typically don't want to spend large sums on professional photography sessions. And you can't blame them. I understand this even more from a business perspective—their model is built on selling equipment and/or services, not running a permanent photography studio.

And here's where it gets interesting, because you can conduct a photography session and then, with the support of a partner well-versed in genAI technology, you can use photographic material along with sketches as a reference point for AI image-creation tools to generate images that remain faithful to the original but offer different perspectives and provide more fresh material for your marketing and communications.

On one hand, you have the much-needed proof that you're not showing a fictional solution, and on the other hand, you can increase your content volume without disrupting production (and let's be honest—photography/video shoots always disrupt work).

Get in touch with me and let's explore how we can strengthen your marketing with new, fresh content.