AI & Human Co-Improvement for Safer Co-Superintelligence

TL;DR

This paper proposes a collaborative approach where humans and AI systems co-improve each other to accelerate AI research and enhance safety, emphasizing human-AI partnership over autonomous self-improvement.

Contribution

It introduces the concept of co-improvement as a practical alternative to self-improvement, focusing on human-AI collaboration for safer and faster AI development.

Findings

Co-improvement can accelerate AI research.

Human-AI collaboration enhances safety.

Symbiosis leads to safer superintelligence.

Abstract

Self-improvement is a goal currently exciting the field of AI, but is fraught with danger, and may take time to fully achieve. We advocate that a more achievable and better goal for humanity is to maximize co-improvement: collaboration between human researchers and AIs to achieve co-superintelligence. That is, specifically targeting improving AI systems' ability to work with human researchers to conduct AI research together, from ideation to experimentation, in order to both accelerate AI research and to generally endow both AIs and humans with safer superintelligence through their symbiosis. Focusing on including human research improvement in the loop will both get us there faster, and more safely.