The Ethical Failure of Recognition Technologies

Although facial recognition software has become increasingly widespread, practices of data mining, data inputting and algorithm creation used to classify and differentiate individuals have made these technologies fundamentally unethical.

From these photos, you’ll be able to see blatant algorithm flaws and issues of design, some of which have been fixed by tech companies after consumer outrage and others which have the potential to pose as a threat to the wellbeing of societies. Each picture has a clickable link that leads to its respective news article.

• 

• 

• 

• 

---

• 

• 

• 

What if I told you that the turtle was a rifle, the hospital was a drone, and the child on his bike is a wanted criminal?

Well, I can tell you that it’s not actually true, but the issue is that it could be. 

 This is because it is possible to trick image recognition algorithms. For example in the case of the turtle, by subtly tweaking the pattern on its shell, researchers were able to trick a neural network into misidentifying the turtle as a rifle. 

Obviously these results raise concerns about the accuracy of face recognition systems 

These cases are known as an “adversarial examples”.