<aside> 💡 Read Excavating AI: The Politics of Images in Machine Learning Training Sets by Kate Crawford and Trevor Paglen. Consider the following excerpt from the conclusion:
The artist René Magritte completed a painting of a pipe and coupled it with the words “Ceci n’est pas une pipe.” Magritte called the painting La trahison des images, “The Treachery of Images.”
Magritte’s assumption was almost diametrically opposed: that images in and of themselves have, at best, a very unstable relationship to the things seem to represent, one that can be sculpted by whoever has the power to say what a particular image means. For Magritte, the meaning of images is relational, open to contestation. At first blush, Magritte’s painting might seem like a simple semiotic stunt, but the underlying dynamic Magritte underlines in the painting points to a much broader politics of representation and self-representation.
Reflect on the relationship between labels and images in a machine learning image classification dataset? Who has the power to label images and how do those labels and machine learning models trained on them impact society?
</aside>
What does the world that the machine recognizes looks like? The conversion from a qualitative real world to a quantitative digital environment can often lead to misunderstandings, as they have different compositions. Unlike the flexible and context-rich nature of the real world, machines have to translate these ideas into numeric expressions. This is also true for machine learning. While its results may appear similar to human natural language, it is important to note that the way it reads and categorizes people has significant limitations.
“What if the challenge of getting computers to “describe what they see” will always be a problem?”
Source: https://datagen.tech/guides/image-annotation/image-labeling/
Labeling is a useful technique for categorizing objects into specific groups. By analyzing their visual appearance, machines can identify visual similarities and characteristics using numerical data and pixelated image readings. However, machines cannot interpret the context behind the image, which may result in subjects being categorized solely based on their visual features. For example, the machine may not consider the possibility of a man being a woman or having a mixed race, as it relies solely on visual elements.
”Despite the common mythos that AI and the data it draws on are objectively and scientifically classifying the world, everywhere there is politics, ideology, prejudices, and all of the subjective stuff of history. When we survey the most widely used training sets, we find that this is the rule rather than the exception.”
Yes, when trained, machines can “neutrally” read and identify all the ambiguity. They can mathematically tell the percentage of race of people with their interpretation based on detectable characteristics. The machine can classify me as an Asian by more than a 90% chance based on my appearance.
”Regardless of the supposed neutrality of any particular category, the selection of images skews the meaning in ways that are gendered, racialized, ableist, and ageist.”
However, what does it mean to be an Asian? The machine categorizes me in a particular category because I have characteristics that other Asians have. This might not be true if I were an Asian and didn’t have common similarities. These technically unbiased groupings have a risk of alienating and creating stereotypes.
Source: https://time.com/5520558/artificial-intelligence-racial-gender-bias/
Indeed sometimes being objective in every situation might not be the best solution. This is why humans can bring many meanings and interpretations from one subject. We can say and comprehend the different understandings surrounding the same matter because we can weave context and appearance. On top of that, humans also have ideas and emotions. They can help to interact and know the truth beyond the visual status. Invisible context and emotions are everywhere to help us make decisions. Indeed “everywhere there is politics, ideology, prejudices, and all of the subjectivity,” as Kate mentioned in Excavating AI.