Humans in the Loop (Aranya Sahay) India |

Humans in the Loop
Aranya Sahay
India. 2024. 74 min

Nehma, an Adivasi woman from the Oraon tribe, returns to her village with her children twelveyear- old Dhaanu and one-year-old Guntu, after her divorce. She begins to work as a data labeller, training the ‘childlike’ AI models to recognise images.
While Nehma faces the challenge of connecting with Dhaanu – who is biased against village life and forever tempted to flee back to the city she also notices that the AI she works with is adopting harmful human biases, echoing prejudices against her own community.

ARANYA SAHAY

Aranya is an Indian filmmaker based out of Mumbai. He studied Direction at FTII, Pune. He went on to assist Imtiaz Ali and Patrick Graham as an Associate Director & Director’s Assistant.He has directed five short films, of which ‘Songs forBabasaheb’ and ‘Chait’ have screened at Mumbai International Film Festival, Signs Film Festival and Jaffna Film Festival. In 2021, his film ‘Saaya’ was selected as a ‘Film Bazaar Recommends’ project. He was a participantand grant winner in the Museum of Imagined Futures Impact Fellowship Programme in 2023.

DIRECTOR,S NOTE

In 2022 I read an article titled ‘Human Touch’ by Karishma Mehrotra about tribal women in Jharkhand working with artificial intelligence as data annotators. Annotators ‘label’ images, videos and audio in order to improve machine learning algorithms. When exposed to new data, these applications learn and develop by themselves, learning to differentiate between objects and recognise patterns, much like a child.

This provided the inspiration for a civilizational tale—one that explores the interplay between a traditional society and technologies of the future. While researching in Jharkhand, I came across remnants of colonialism – buildings, arches, institutions, and a painful history of imperialism that destroyed indigenous communities, dismissing their knowledge as “primitive.” The rise of AI, largely shaped by Western perspectives, threatens to replicate this exclusion.

This phenomenon, known as “AI Bias,” perpetuates human prejudices through limited and biased data that is used to train machine learning models, eliminating the identities and cultural knowledge of the very people who are recruited to label this data.

My mother’s work as a researcher in tribal regions of India left an indelible mark on me. It has long inspired me to tell a story set in Jharkhand. There is place no better than this to situate an indigenous story set against the backdrop of a world that is rapidly changing through AI.