Colorizing With AI

Artificial intelligence is becoming increasingly popular, especially with how easily it can be accessed in programs like ChatGPT. AI increases efficiency and can automize tasks for people. While AI is a useful and powerful tool, it also raises ethical concerns. It is so easy for AI to create inaccurate information such as fake news or the color of a photo. The process of colorizing images with artificial intelligence has the potential to introduce bias into the photos. An algorithm can’t know what the color of a photo is supposed to be. Drimmer explains,

How absurd to think that black-and-white photographs from 100 years ago would produce colors in the same way that digital photographs do now. And yet, this is exactly what AI assisted colorization does.

Drimmer, Sonja. “How AI Is Hijacking Art History.” The Conversation, The Conversation, 1 Nov. 2021, theconversation.com/how-ai-is-hijacking-art-history-170691. 

Drimmer’s concerns are illustrated in the image I chose to colorize from our class exercise using DeOldify AI. There is no way to know if the colors are representative, and they can stem from bias from the creator. Certain aspects of the photos are clearly inaccurate, like the unusual color of the track.

People may take information of photos produced by AI as factual, which is dangerous. AI may be able to successfully complete tasks, but it still can’t replace people or the work that they do. Drimmer later explains just how scary AI can be, saying:

When AI gets attention for recovering lost works of art, it makes the technology sound a lot less scary than when it garners headlines for creating deep fakes that falsify politicians’ speech or for using facial recognition for authoritarian surveillance.

Drimmer, Sonja. “How AI Is Hijacking Art History.” The Conversation, The Conversation, 1 Nov. 2021, theconversation.com/how-ai-is-hijacking-art-history-170691. 

While the colorization of images isn’t harmful in itself, AI still has the potential to be harmful. It is scary the degree to which it can replace humans. I remember reading about a new AI program where you upload images and texts from a person and you can chat with an AI version of them. In the article, it demonstrated a child using this program to talk to their grandmother they never met. While in theory it is an interesting idea, this child shouldn’t have their entire perception of a person based on AI. With this program and the colorization of images, there is no way to know what inaccuracies and biases are introduced.