Looking to the brain to improve energy efficiency of AI

Clinical Snapshot

Key Findings

• Effect Size: Not applicable; no original quantitative data are generated. The review notes that biological brains perform lifelong learning and adaptive reasoning using 'orders of magnitude less energy' than large-scale AI systems, but this claim is drawn from cited primary literature rather than original analysis

• Primary Outcome: Identification of three domains of biologically informed principles — architectural constraints, signalling strategies, and learning algorithms — proposed to guide the development of more energy-efficient artificial intelligence systems

• Nnt Or Sensitivity: Not applicable; this is a theoretical review. The key conceptual finding is that biological brains maintain remarkably stable energy usage across heterogeneous cognitive modes, suggesting that maximising the utility of 'rest-like' background processes may be a key mechanism for minimising active processing energy costs — a principle proposed as translatable to AI system design

• Confidence Interval: Not applicable; no statistical estimates are produced

Clinical Application

The conceptual framework is intellectually accessible and well-organised, but practical implementation requires substantial interdisciplinary engineering effort. No ready-to-implement clinical tools or protocols are proposed. Feasibility of translating biological principles into deployable neuromorphic AI systems remains a medium-to-long-term research challenge In the Australian healthcare context, this review is indirectly relevant to several emerging priorities. The Australian Digital Health Agency and state health departments are increasingly deploying AI-assisted clinical tools (e.g., diagnostic imaging AI, clinical decision support). The energy and infrastructure costs of these systems are a growing concern for hospital sustainability targets aligned with Australia's net-zero commitments. The Therapeutic Goods Administration (TGA) has begun developing regulatory frameworks for AI as a medical device (Software as a Medical Device, SaMD), and energy efficiency is not yet a formal regulatory criterion — this review implicitly supports the case for including sustainability metrics in future TGA guidance. The RACGP has no current guidelines directly addressing AI energy efficiency, but the principles discussed are relevant to general practice adoption of AI tools. PBS implications are not directly applicable. Australian research funding bodies (NHMRC, ARC) may find the interdisciplinary framework relevant to priority-setting in biomedical AI research grants. Not directly applicable to a patient population. Relevant stakeholders include AI researchers, computational neuroscientists, health technology developers, hospital IT infrastructure planners, and health policymakers concerned with the environmental and economic sustainability of AI-driven clinical decision support systems

Abstract

Modern artificial intelligence (AI) systems have achieved remarkable capabilities, but at an extraordinary energy cost. Training and running large-scale models can consume vast resources, posing environmental, economic, and societal challenges. In contrast, biological brains perform lifelong learning, adaptive control, and flexible reasoning using orders of magnitude less energy for learning and adaptation over a lifetime. What accounts for this difference - and how can it guide future AI development? In this review, we identify key biological principles that support energy-efficient capacities in biological brains, and consider how they might inform the design of more sustainable artificial systems. We organize our analysis around three domains: architectural constraints, signaling strategies, and learning algorithms. In each domain, we discuss concrete observations from biology, from cell to circuit to cognitive level, and describe how current and emerging AI systems mirror or diverge from these motifs. One striking feature of biological energy optimization is often overlooked: that brains are remarkably stable in their energy usage across heterogeneous modes, suggesting they may minimize energy needs during active environmental processing through maximizing the utility of 'rest-like' background processes. Overall, rather than advocating for biomimicry for its own sake, we argue for biologically informed engineering. Understanding how natural systems minimize energetic cost while maximizing flexibility may help us build AI that is not only powerful, but also efficient, equitable, and environmentally responsible.

References

1. 1.Peters, M. A. K., Yoo, M. S. B., Klincewicz, M., Toyoizumi, T., Richards, B., Webb, T., & Lau, H. (2026). Looking to the brain to improve energy efficiency of AI. Current Biology, advance online publication. https://doi.org/10.1016/j.cub.2026.06.020