# From AGI to ASI

**Authors:** Tim Genewein, Matija Franklin, Alexander Lerchner, Laurent Orseau, Samuel Albanie, Adam Bales, Cole Wyeth, Stephanie Chan, Iason Gabriel, Joel Z. Leibo, Allan Dafoe, Marcus Hutter, Thore Graepel, Shane Legg  
**arXiv:** [2606.12683](https://arxiv.org/abs/2606.12683)  
**Date:** 2026-06-10  
**Summary page:** https://www.summarizepaper.com/en/paper/2606.12683v1/

## Key points

- Pursuit of human-level artificial general intelligence (AGI) has become a tangible target for major AI organizations
- Implications of achieving AGI prompt complex questions and considerations for society
- Evolution of AI beyond AGI towards artificial superintelligence (ASI) is explored in the report
- Four potential pathways to ASI: scaling AGI, paradigm shifts in AI development, recursive improvement processes, and emergence through large-scale multi-agent systems
- Possible obstacles and bottlenecks along pathways highlight the need to address key research questions
- Uncertainties around ASI progress suggest advancements in AI could accelerate in the future
- Transformative changes driven by breakthroughs across scientific and technological domains enabled by AI progress are expected instead of a singular transformative event triggered by human-level AGI introduction
- Collaborative interdisciplinary effort on a global scale is necessary to prepare for the evolving landscape shaped by advancing towards ASI
- Proactive engagement with emerging technologies is crucial to anticipate and navigate potential societal impacts positively

## Summary

In recent years, the pursuit of human-level artificial general intelligence (AGI) has shifted from speculative to a tangible target for major AI organizations. The potential achievement of AGI holds significant implications for society, prompting complex questions and considerations for the future. This report delves into the evolution of AI beyond AGI towards artificial superintelligence (ASI), which surpasses human cognitive capabilities. The concept of Universal AI serves as a theoretical foundation for exploring the transition from AGI to ASI. The report identifies four potential pathways leading to ASI: scaling AGI, paradigm shifts in AI development, recursive improvement processes, and the emergence of ASI through large-scale multi-agent systems. It also highlights possible obstacles and bottlenecks along these pathways, emphasizing the need to address key research questions to assess their impact. Given the uncertainties surrounding ASI progress, there is a possibility that advancements in AI could accelerate in the coming years. This challenges the notion of a singular transformative event triggered by the introduction of human-level AGI into society. Instead, it suggests a series of transformative changes driven by breakthroughs across various scientific and technological domains enabled by AI progress. Preparing for this evolving landscape requires a collaborative interdisciplinary effort on a global scale. By addressing critical issues and fostering innovation in AI development, researchers can better anticipate and navigate the potential societal impacts of advancing towards ASI. Ultimately, this report underscores the importance of proactive engagement with emerging technologies to shape a future where artificial intelligence contributes positively to human progress and well-being.