Want to hear something chilling? Amazon can sense your fear.

Maybe I should have saved this post until Halloween, because if you’re looking for a scare, look no further than image recognition technology. Privacy concerns around facial recognition technology in particular have come up a lot in the news lately. Some politicians are calling to ban it altogether. I didn’t know much about it, so I took a look around.

Let’s focus on Rekognition, Amazon’s facial recognition technology. On a basic level, the technology works by plotting the characteristics of your face. But we’ve moved way past just recognizing features. It can recognize demographics like gender and age. Amazon touts its ability to guess if you’re happy, sad, angry, surprised, disgusted, calm, or confused. The ability to detect fear was added this summer. It works on still images as well as video. And general image recognition is important here too because it contributes to understanding context. Think: "That's an image of Bob. He's happy. There's a pool below him and a diving board being him. Bob's jumping. Bob's jumping into a pool."

From a marketing perspective, emotive and demographic attributes can be used to show more customized advertising. It’s been used to gauge customer reaction to new products. Amazon Redshift can process data from Rekogniton and report on trends such as in store locations and similar scenarios. 

Or maybe you've used Snapchat's image recognition feature, which was rolled out in September 2018 in partnership with Amazon. Simply hold down on the screen and Snapchat can visually identify products and will provide you with an Amazon link to purchase said product. I'm impressed by how well this works. When I hold down on an image of my dog, not only does Snapchat recognize there is a dog in the picture, but they suggest that I might like to buy a puppy starter kit and dog food. I also thought it was interesting that when I tried this exercise with my other dog the first suggestion was for a Patrick Swayze movie called Black Dog. Ok, I guess it's not perfect yet. 

Other suggested applications for image recognition technology start at creepy then quickly veer off into more sinister territory. For example, maybe it’d be useful to get a notification if your dog jumps up on the white sofa while you’re at work (even the dog can’t catch a break). Or, let your pastor know if you skipped church and that when you do come you like to steal toilet paper from the bathroom (for shame!). Wouldn’t it be nice for an HR manager to have an emotion detecting tool that shows if you’re being dishonest in an interview or during pay negotiation? Maybe the authorities could use it to know if you forgot to clean up after your dog, so they can let everyone know what kind of person you really are. 

China’s gotten a lot of attention for this last point. But it might surprise you to know how common the technology is in the US. While the Asia Pacific is the fastest growing market for facial recognition, North America remains the largest. Law enforcement in particular is a major user.

And despite all the concerns and calls to outlaw it, I’m betting it’s here to stay. The technology just offers too many perceived conveniences and the data it collects is just too valuable for it to go away. The facial recognition market is valued at USD 4.51B in 2018 and is expected to reach a value of USD 9.06B by 2024 (at a growth rate of 12.5%). And since there is such a range of possible applications for this technology there’s no way to know all of the implications it will have.