# A Framework for Searching for General Artificial Intelligence

**Authors:** Marek Rosa, Jan Feyereisl, The GoodAI Collective  
**arXiv:** [1611.00685](https://arxiv.org/abs/1611.00685)  
**Date:** 2016-11-02  
**Summary page:** https://www.summarizepaper.com/en/paper/1611.00685v1/

## Key points

- Lack of unified approaches in building generally intelligent machines
- Current AI research operates within narrow fields without considering the bigger picture
- Proposal of principles to guide the development of general artificial intelligence
- Intelligence as a tool for searching for general solutions to problems
- Definition of intelligence as the ability to acquire skills that narrow down, diversify, and steer the search
- Importance of studying, measuring, and testing various skills and abilities required by human-level intelligent machines
- Implementation agnostic approach with an analytic, systematic, and scalable approach to generating hypotheses
- Framework serves as a stepping stone in bringing together definitions, highlighting open problems, connecting researchers, and unifying AI development projects
- Comprehensive overview of principles guiding the development of general artificial intelligence
- Suggestions for further research in this field
- Holistic approach and addressing key challenges advance progress towards achieving truly intelligent machines
- Contribution to understanding and progress towards achieving truly intelligent machines

## Summary

In their paper titled "A Framework for Searching for General Artificial Intelligence," Marek Rosa, Jan Feyereisl, and The GoodAI Collective address the significant lack of unified approaches in building generally intelligent machines. They highlight that current artificial intelligence research often operates within narrow fields of focus without considering the bigger picture. To bridge this gap, the authors propose principles that guide the development of general artificial intelligence. The central idea presented in this framework is that intelligence serves as a tool for searching for general solutions to problems. The authors define intelligence as the ability to acquire skills that narrow down this search, diversify it, and steer it towards more promising areas. They emphasize the importance of studying, measuring, and testing various skills and abilities required by human-level intelligent machines. The document aims to be implementation agnostic while providing an analytic, systematic, and scalable approach to generating hypotheses necessary for achieving general artificial intelligence. The authors believe that this framework serves as a crucial stepping stone in bringing together definitions, highlighting open problems, connecting researchers willing to collaborate on AI development projects and unifying what they consider to be one of the most significant searches of this century. Overall, the paper provides a comprehensive overview of the principles guiding the development of general artificial intelligence and offers suggestions for further research in this field. By emphasizing a holistic approach and addressing key challenges in AI development progress towards achieving truly intelligent machines is advanced. This framework contributes significantly to our understanding and progress towards achieving truly intelligent machines.