From Social Feeds to Firepower: The Black Box We Can’t Ignore
by Utkarsh Bajpai
The United Nations (UN) International Youth Day is celebrated every year on 12 August to bring global attention to the voices, actions, and initiatives of young people. In honour of this important day, members of the Stop Killer Robots Youth Network shared their thoughts on the themes of autonomous weapons systems, digital dehumansiation, and youth in disarmament. Disclaimer: The blogs in this series do not necessarily constitute the opinions of Stop Killer Robots, nor should they be considered the opinions and views of all Stop Killer Robots members. The views expressed here are solely those of the author and do not reflect the positions of the author’s employer, academic institution, or any other affiliated organisation.
Young people today are navigating a world increasingly governed by artificial intelligence (AI) systems. Rather than relying on traditional media like newspapers or television, they engage with information and express themselves through algorithm-driven platforms such as Instagram, TikTok, and Twitter. These are not just communication apps but environments that shape identity, opinion, and knowledge. Over time, users build a carefully curated feed: liking, sharing, and posting content until their digital presence feels uniquely personal. But what happens when that digital presence is removed without warning or explanation? Recently, thousands of Instagram users experienced this firsthand when their accounts were suspended without warning, despite following platform guidelines. Meta later attributed this wave of suspensions to an error in its automated moderation system, an AI tool designed to detect and remove harmful content, but could provide no explanation for why specific users were misclassified and banned, leaving affected users with no clarity or recourse.
This incident reflects a broader issue in how AI systems operate and is not isolated to social media filters. The core problem lies in the lack of explainability in AI-based decision-making. Without explainability, or, the ability to understand and explain how decisions are made, users are left vulnerable to opaque “black box” AI models, which provide little accountability when things go wrong. Another AI tool that is gaining widespread usage for writing, research, and communication among youth are large language models such as ChatGPT. These models can generate text and function as AI assistants, yet they frequently generate fabricated or misleading information, so-called “hallucinations”. In 2025, a U.S. lawyer cited multiple fictitious cases generated by an AI assistant, highlighting the risks of relying on systems that produce plausible but unverifiable content. These issues are not mere bugs; they stem from the architecture of current AI systems, which lack mechanisms for transparent reasoning. Even AI researchers acknowledge that they often cannot fully explain how these models arrive at their outputs. Although there has been research in this direction, attempts to interpret the inner workings of large AI models remain elusive.
When AI is applied to military contexts, the consequences escalate dramatically. Autonomous Weapons Systems (AWS), colloquially referred to as “killer robots”, inherit many of the same limitations seen in civilian AI, including a lack of transparency and explainability. When errors such as misclassification occur in life-or-death scenarios, accountability becomes fragmented across developers, military operators, and the AI systems themselves, making it nearly impossible to determine who is responsible. These concerns are no longer theoretical. In ongoing conflict regions, autonomous systems such as drones are reportedly identifying and striking targets without direct human supervision, relying on classification systems akin to those used in content moderation, and sometimes incorporating generative models for navigation and decision-making. Meanwhile, AI companies are actively developing large language models to support battlefield decision-making. The integration of opaque AI into military infrastructure raises urgent concerns about safety, oversight, and ethical governance.
At its core, this challenge reflects growing digital dehumanisation, or the reduction of individuals to data points processed by automated systems operating without human context or accountability. Although these systems learn from human-annotated data, their decision-making often lacks transparent reasoning and direct human judgment, leading to fragmented accountability across developers, operators, institutions, and algorithms. Users may find their accounts banned without explanation, encounter fabricated information from chatbots, or, in more serious contexts, witness autonomous systems misclassifying targets that could cause civilian harm without clear traceability. These issues mirror documented technical failures in AI research, such as classification errors, data biases, and overfitting. Addressing them requires both technical advances in explainable AI architectures and robust governance frameworks that ensure transparency, verification, and accountability.
Explainability is not just a technical aspiration; it is a democratic necessity. Without the ability to inspect, audit, or challenge algorithmic decisions, we surrender authority to systems we cannot question. This is especially concerning for young people, who are both the most engaged users of AI technologies and the generation that will bear the long-term implications of these technologies. On International Youth Day, young people must assert their right to question the systems shaping their lives, including the algorithms that decide what they see, say, and can do. As AI becomes increasingly embedded across dual-use contexts, spanning civilian platforms and military operations, demanding better performance is not enough; we must insist on transparency, accountability, and explainability by design. Until these mechanisms are in place, delegating lethal decision-making to “black box” AI systems poses an unacceptable risk to responsible societies.
Utkarsh Bajpai is a PhD researcher in Robotics and a Marie Skłodowska-Curie Fellow at Georgia Tech-CNRS. His work focuses on developing reliable AI architectures for robotic systems aimed at societal benefit. He is passionate about the technological advancement and ethical use of autonomous technologies.
