# Cancermorphic Computing Toward Multilevel Machine Intelligence

**Authors:** Rosalia Moreddu, Michael Levin  
**arXiv:** [2503.12743](https://arxiv.org/abs/2503.12743)  
**Date:** 2025-03-17  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2503.12743v1/

## Key points

- Cancer-inspired computing, or cancermorphic computing, draws from cancer cells' adaptive, resilient, and evolutionary strategies to design computational systems for dynamic environments.
- The approach emulates survival tactics of cancer cells such as somatic mutation, metastasis, angiogenesis, and immune evasion to enhance fault tolerance and cybersecurity in computational systems.
- Focuses on features like decentralized propagation and resource optimization rather than traditional bioinspired approaches like evolutionary and neuromorphic architectures.
- Incorporates randomness-based algorithms to enhance computing capabilities by leveraging concepts of multilevel intelligence and context-driven mutation inspired by cancer mechanisms.
- Challenges include effectively emulating cancer survival strategies while ensuring system stability and alignment with human objectives.
- Need for a conceptual shift towards recognizing the potential for cancer-inspired mechanisms to inform the development of powerful artificial systems through interdisciplinary discussions and research efforts.

## Summary

The concept of cancer-inspired computing, also known as cancermorphic computing, introduces a paradigm that draws from the adaptive, resilient, and evolutionary strategies of cancer cells to design computational systems capable of thriving in dynamic, adversarial, or resource-constrained environments. This innovative approach aims to address crucial bottlenecks in computing architectures by emulating the unique survival tactics of cancer cells such as somatic mutation, metastasis, angiogenesis, and immune evasion. Unlike traditional bioinspired approaches like evolutionary and neuromorphic architectures, cancer-inspired computing focuses on features like decentralized propagation and resource optimization to enhance fault tolerance and cybersecurity in computational systems. While the chaotic growth of cancer is typically viewed as uncontrollable in biology, this concept suggests that randomness-based algorithms can be successfully integrated into computing architectures to enhance their capabilities. By incorporating concepts of multilevel intelligence and context-driven mutation inspired by cancer mechanisms, this approach seeks to overcome limitations seen in current neuromorphic approaches while leveraging the randomness observed in chaotic systems. However, challenges remain in effectively emulating cancer survival strategies while ensuring system stability and alignment with human objectives. The prevailing view of cancer as a genetic disorder rather than a model of resilience poses a conceptual challenge that needs to be addressed. Additionally, theoretical challenges arise from the perception that cancer represents a failure of biological regulation rather than a source of valuable insights for constructing capable agents. To advance this novel concept further, there is a need for a conceptual shift towards recognizing the potential for cancer-inspired mechanisms to inform the development of powerful and resilient artificial systems. By exploring new directions and overcoming existing challenges through interdisciplinary discussions and research efforts, the field of  holds promise for revolutionizing computational paradigms and unlocking new possibilities for machine intelligence.