Humans in the Loop
Movie Review
Two weeks ago, I got to view the Premiere of ‘Humans in the Loop’ Directed by Aranya Sahay. I was excited that an Indian director had explored the topic of Artificial Intelligence and was keen to listen to his point of view.
‘Humans in the loop’ raises an important question: “What are the biases that we would pass on to Artificial Intelligence?”
The context chosen for this is interesting. The stage is set in contemporary India. A remote tribal village in rural Jharkhand. A small group of indigenous people are being trained to label data, which is then used to train AI models abroad. The entire data labelling task is outsourced to India.
Nehma is an indigenous woman who is back in her village after separating from her husband. She contemplates life ahead with her teenage daughter and her newborn son. She is eager to gain the data labelling work and support her children.
Nehma has a tacit understanding of the forest, animals, trees and insects. This is the knowledge that has been passed on to her from her ancestors. She can spot the subtle differences between seemingly similar-looking roots or insects. She has imbibed this knowledge from having spent her formative years in the forest and observing nature closely.
Nehma struggles to keep her teen daughter, Dhaanu, with her. The teenager doesn’t like the shift from the urban city to the rural forest. Dhaanu has trouble adjusting to her village school and friends. She cannot seem to figure her way out of it. Nehma wants to teach her teenage daughter their adivasi way of life.
Nehma wants to teach AI the right way. She wants to pass on her deeper understanding to train the AI model as much as she wants her children to learn the right ways of the world. She is rebuffed and chastized for trying to go against the client’s stated requirement.
Get Ravi Pudi’s stories in your inbox
Join Medium for free to get updates from this writer.
https://youtu.be/GxKmuCENXK4?si=MvDfbeMEleksxL89
A badly trained AI model is likely to harm the world. It could used to eliminate innocent insects wrongly labelled as harmful.
But what happens when we as a society chose to ignore the wisdom and knowledge of under represented communities?
Nehma is alarmed when AI renders an inaccurate image of an Adivasi woman. She realizes that the data that AI has been fed is biased and represents what the West sees. But she learns that with careful attention and proper inputs she can train AI to understand and represent the marginalized as well.
The movie succeeds in raising the right questions.
Are we as a society going to focus on the right models?
How much effort would we put to ensure such intelligence learns to ‘listen and feel deeply’?
