Perspectives

Although my day-to-day work centers on ML engineering and physics research, I’ve long been drawn to fundamental questions about how intelligence operates and what ethical responsibility might entail – for both humans and machines. These reflections are purely personal and independent and do not represent the views of any institution. They build on my physics-driven pursuit of simple yet universal models, draw on state-of-the-art insights in RL and continual learning, and are informed by years of personal exploration in psychology, cognition, and philosophy.

​

As AI embeds deeper into society, concepts once deemed purely philosophical – like consciousness or moral responsibility – are proving urgent, practical, and impactful. Here, I propose two evolving frameworks – ADAPT and CURE – whose core lies in a novel (to the best of my knowledge), physics-inspired insight: that consciousness can be treated as the “time derivative” of intelligence (akin to velocity being the derivative of position), and that ethical responsibility can be reframed around an agent’s “capacity to retrain.” While many components of these frameworks are grounded in existing research from cognitive science, RL, and ethics, the unified framing across biological and artificial systems is my own personal contribution. I see these ideas as complementary to my technical work, offering a pragmatic, long-term vision for adaptive, ethically grounded AI systems.