Using AI Responsibly

A naturally-occurring neural network's attempt at reproducing a known meme image.

The problem with today's AI/ML craze is that companies are too hot for the dollar bills associated with the buzz. They try too hard to make AI the star feature of their product.

That's the wrong way to use it.

It's understandable. Companies exist to make money, after all. Having a cool tech demo could be the difference between success and failure.

Setting aside the ethical and legal considerations, there are two main reasons why it's dangerous to put AI front and centre in your business:

  1. It's not as good as we've been told it is.
  2. There's no inherent value.

With this technology expanding at a geometric rate, your AI/ML solution will become outdated almost immediately. Competitors will be able to spin up more powerful replacements with little effort, which you'll have to counter - or your product's value will plummet.

Building a company around a whizzbang AI/ML stage show is not sustainable. It's an arms race.

Instead, focus on using AI/ML tech where it shines: boosting your product's innate value. Make it the cherry on top of an excellent customer experience.

Adobe has helpfully provided two examples for this discussion, one on each side of the argument. Thanks, Adobe!

In one corner, we've got Adobe Firefly.

This product helps users generate imagery from text prompts, or modify existing images according to other instructions. This sort of generative art has led to some neat results, but also to some hilariously disastrous ones.

In the other corner, we have Lightroom's "AI Noise Reduction" feature.

Passing your photo through this filter results in a more effective noise reduction. It's not perfect, it's not going to magically transform your photos into tack-sharp focus - but I've yet to see any major issues with it.

One of these tools invents things from whole cloth.

The other is a nuanced refinement of an existing tool.

One of them takes source data from the creativity of thousands of photographers and stands a very real chance of being party to the destruction of livelihoods.

The other looks at thousands of camera exposures and uses that analysis to subtly enhance the final product by removing sensor defects and lens aberrations.

One of them is a surefire strategy for capturing press attention and (in the case of startups) fresh investment.

The other is a quality of life improvement for photographers and fellow creatives.

Don't be the company that puts AI gimmickry at the forefront of your application & advertising. Don't use it to remove the human element from the creative process. Don't cement yourself into an unwinnable arms race as your company spirals into valueless oblivion.

Use it the way it should be used: making lives easier and more rewarding.

The only thing that should be front and centre is your customers.

Deliver value first.

Published: July 16, 2023

Categories: opinion

Tags: opinion, photography, machine-learning