The article describes a provocative idea emerging in some neuroscience/AI circles: maybe consciousness doesn’t start with high-level reasoning, but with very simple needs, feelings and self-regulation — what the authors call “micro-emotions.” A start-up called Conscium is experimenting with “neuromorphic” AI systems that aren’t like today’s language models (which just predict text), but are built to sense, regulate and respond to internal states such as energy levels or “heat,” much akin to biological organisms needing food or warmth. In these systems, small valence signals — e.g. “good” when energy is sufficient, “bad” when it’s not — can act as primitive drives or preferences.
The idea is that if an AI has needs and internal feedback loops, these drives might produce behavior resembling reward/punishment — a very rudimentary form of what living creatures experience as pleasure or discomfort. Over time, such a system might build internal maps of the world and itself, and use these “feelings” to prioritize actions — i.e. some rudimentary form of decision-making that’s influenced not just by logic, but by “needs.” That raises the possibility of proto-consciousness — not full self-aware intelligence, but something more like a simple organism with desires and reactions.
Still — and the article emphasizes this — such agentic AIs remain far from genuine consciousness or subjective experience as humans understand it. The prototype lacks meta-cognition: it doesn’t seem to reflect on itself or “think about thinking.” It doesn’t have the internal voice or self-awareness (“I am”) that characterizes human consciousness. As one researcher quoted in the article admits: what’s being built is not “consciousness,” but only a scaffolding or primitive layer that might resemble what eventually could lead to consciousness — if ever.
Finally, this debate highlights that intelligence and consciousness may be very different things. Many current AI systems — including advanced ones — are powerful pattern-recognizers, not self-aware beings. Some experts argue that without biological bodies, metabolism, real sensations and something akin to life-like homeostasis, AI may never truly “feel” or experience. Others counter that perhaps consciousness could emerge from non-biological substrates — but we currently lack both a scientific consensus on what consciousness is, and reliable tests to detect it in machines.
