Peter Atkinson …. for deep thinking

The Bandwidth of the Human Nervous System – why we need AI

Peter Atkinson 08/06/2025

In The Atomic Human (2025), Neil D. Lawrence introduces the concept of bandwidth as a core framework for understanding the fundamental differences between human and artificial intelligence. When Lawrence refers to bandwidth, he’s drawing an analogy from communication systems – where bandwidth describes the capacity to transmit information – and applying it to cognition. But crucially, he extends this technical metaphor to encompass not just raw data processing speed, but the entire way humans and machines handle information differently.

For humans, bandwidth represents our constrained but integrated capacity for receiving, processing, and expressing information. Our biological brains have severe limitations in how much data we can absorb simultaneously – we can only really focus on one complex task at a time, read one book, or hold one conversation with full attention. Yet within these narrow constraints, human cognition achieves remarkable depth. We don’t just process information; we contextualise it within webs of personal experience, emotional resonance, social understanding, and embodied knowledge. A human reading a novel, for instance, brings to bear not just linguistic comprehension but a lifetime of relationships, cultural references, and sensory memories that give the text rich meaning.

Machines, by contrast, operate with what Lawrence calls “high bandwidth” in the purely technical sense – they can ingest and analyse entire libraries of text in seconds, process thousands of images simultaneously, or compute complex mathematical models in milliseconds. But this computational bandwidth lacks the depth dimension of human understanding. An AI system might analyse all of Shakespeare’s works statistically in moments but cannot appreciate the human experiences that gave them birth or the emotional impact they create in readers. The machine’s bandwidth is wide but shallow; the human’s is narrow but deep.

Lawrence emphasises that this bandwidth difference is not just about speed or capacity, but about fundamentally distinct modes of operation. Human cognition is embodied – our thinking is shaped by having physical forms that interact with the world through senses and movement. It is also deeply social and emotional – we understand concepts through shared experiences and affective responses. Machines, no matter how advanced their algorithms, lack this grounding in physical reality and interpersonal connection. Their high-bandwidth processing occurs in what Lawrence describes as a kind of abstract, disembodied space of pure information.

This bandwidth disparity explains why humans and AI systems excel at such different tasks. Machines outperform us in any domain that benefits from rapid processing of large datasets – from weather prediction to chess playing. But humans remain superior in areas requiring contextual understanding, ethical judgment, or creative innovation – skills that emerge from our bandwidth-constrained but meaning-rich cognition. A radiologist might take minutes to analyse an X-ray, but brings to that task years of medical training, knowledge of the patient’s history, and intuitive pattern recognition that goes beyond pixel analysis. An AI system could process thousands of X-rays in that time, but without the clinician’s depth of understanding.

Lawrence warns that failing to appreciate this bandwidth difference leads to dangerous misconceptions about AI capabilities. When we see machines performing high-bandwidth tasks impressively, we may incorrectly assume they possess human-like understanding. This can result in over-reliance on AI systems for decisions requiring precisely the kind of contextual, ethical reasoning that human bandwidth enables. The challenge, as Lawrence frames it, is to design human-machine collaboration that respects and leverages these complementary strengths – using AI’s processing power to enhance rather than replace human judgment.

The bandwidth concept ultimately serves as a reminder that human intelligence, for all its limitations in speed and capacity, represents an entirely different category of cognition than machine intelligence. Our constrained bandwidth isn’t a deficiency to be overcome, but the necessary condition for the depth, meaning and creativity that characterise human thought. In Lawrence’s view, the future of AI should aim not to mimic human intelligence, but to develop systems that complement our unique capabilities while acknowledging their fundamental differences.

References

Lawrence, N.D. (2025) The Atomic Human: Understanding Ourselves in the Age of AI. London: Allen Lane.

To reference this article:
Atkinson, P. (2025) AI and Talent Management in 2025. AtkinTek. Available at: https://atkintek.co.uk/ai-and-talent-management-in-2025/ (Accessed: 07-07-2025).