Opinion AI and a child cannot be compared. Why ‘Humans in the Loop’ unsettles me

In The Child and the Curriculum, John Dewey, an American philosopher and educator, describes children as active learners whose present experiences are dynamic and consist of a “developing force”, essential to learning. According to him, learning must involve the establishment of a connection between the mind and what is learnt. Without an “organic connection with what the child has already seen and felt and loved,” the child would lose interest, and learning would become burdensome. Learning is, hence, not merely filling a child’s mind with external inputs but engaging, discovering, and actively inquiring.

The primacy of a child’s experiences and exploratory instincts for learning makes it incomparable to the mechanistic process of feeding data into AI. To say the opposite is to wrongly attribute consciousness of experience to the AI model. This is what the film does in its unwarranted comparison. It elevates AI to the stature of child-like innocence and ability to change, while, at the same time, it patronises the way a child learns and relegates complete control of her learning to external authorities.

The film is also replete with contradictions in its own comparison between a child and AI. In its portrayal of Nehma’s childhood, it shows her in the wilderness of the forest as a young girl, interacting with a porcupine and wandering freely in nature. Years later, her teenage daughter, Dhannu, also ventures into the forest and struggles to find her way out. In the process, she learns to navigate her way around with the help of a porcupine, throwing light on how children learn through active engagement with the environment. In these scenes, the film captures how Adivasi children learn about nature in an ontological unity and oneness with animals, insects, plants, trees, and rivers. However, the very next moment, it gets into the AI-child comparison, weakening the aspect of Adivasi life where children grow up and learn in intimacy and closeness with their environment.

The film attempts to delve into complex epistemological issues by debunking the biases in AI models due to the hegemony of knowledge and representation largely borrowed from the dominant Western framework. It depicts the significance of local knowledge and indigenous perspectives and calls for its inclusion in the otherwise one-sided, homogenising information being fed into the AI model. However, it does not raise any serious question about AI itself, or the violation of privacy, as every human move is monitored, labelled and fed into AI models. It never challenges the inherent character of AI as a universalising and standardising tool. The medium itself escapes scrutiny.

What is more striking is that in raising questions about knowledge and representation and comparing AI with children, the film does not take into account how humans learn. It undermines the natural curiosity and sense of wonder inherent in a child by describing it as akin to clay, capable of being moulded as per one’s desire. In this, a child is deprived of any agency in the process of her own learning. A disturbing analogy equating learning with training reduces the process of education to a trivial transfer of information. It assumes that a child is fed information externally and her biases can be removed by mere adult intervention, and these processes can be replicated with AI. Just like AI, a child is not an active participant in her own learning, only a recipient. In the attempt to debunk the Western biases in AI, the film falls into the deadly trap of removing humanity from a child and attributing too much of it to an entity that has “artificial” in the name itself.

The writer is a researcher at Ca’ Foscari University of Venice