Growing Recursive Self-Improvers

Research into the capability of recursive self-improvement typically only considers pairs of \(\langle \)agent, self-modification candidate\(\rangle \), and asks whether the agent can determine/prove if the self-modification is beneficial and safe. But this leaves out the much more important question of how to come up with a potential self-modification in the first place, as well as how to build an AI system capable of evaluating one. Here we introduce a novel class of AI systems, called experience-based AI (expai), which trivializes the search for beneficial and safe self-modifications. Instead of distracting us with proof-theoretical issues, expai systems force us to consider their education in order to control a system’s growth towards a robust and trustworthy, benevolent and well-behaved agent. We discuss what a practical instance of expai looks like and build towards a “test theory” that allows us to gauge an agent’s level of understanding of educational material.

[1]  Jürgen Schmidhuber,et al.  Developmental robotics, optimal artificial curiosity, creativity, music, and the fine arts , 2006, Connect. Sci..

[2]  Kristinn R. Thórisson,et al.  Achieving Artificial General Intelligence Through Peewee Granularity , 2009 .

[3]  B. Bloom,et al.  Taxonomy of Educational Objectives. Handbook I: Cognitive Domain , 1966 .

[4]  D R Krathwohl,et al.  A REVISION OF BLOOMS TAXONOMY: AN OVERVIEW , 2002 .

[5]  Kristinn R. Thórisson,et al.  Why Artificial Intelligence Needs a Task Theory - And What It Might Look Like , 2016, AGI.

[6]  Kristinn R. Thórisson,et al.  Self-Programming: Operationalizing Autonomy , 2009 .

[7]  Kristinn R. Thórisson,et al.  About Understanding , 2016, AGI.

[8]  Benja Fallenstein,et al.  Problems of Self-reference in Self-improving Space-Time Embedded Intelligence , 2014, AGI.

[9]  D. Krathwohl A Revision of Bloom's Taxonomy: An Overview , 2002 .

[10]  Eliezer,et al.  Tiling Agents for Self-Modifying AI , and the Löbian Obstacle * , 2013 .

[11]  Jürgen Schmidhuber,et al.  Anytime Bounded Rationality , 2015, AGI.

[12]  Jürgen Schmidhuber,et al.  Bounded Seed-AGI , 2014, AGI.

[13]  Joan Bruna,et al.  Intriguing properties of neural networks , 2013, ICLR.

[14]  Jason Yosinski,et al.  Deep neural networks are easily fooled: High confidence predictions for unrecognizable images , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Benjamin S. Bloom,et al.  Taxonomy of Educational Objectives: The Classification of Educational Goals. , 1957 .

[16]  Jürgen Schmidhuber,et al.  Gödel Machines: Fully Self-referential Optimal Universal Self-improvers , 2007, Artificial General Intelligence.

[17]  Jürgen Schmidhuber,et al.  Resource-Bounded Machines are Motivated to be Effective, Efficient, and Curious , 2013, AGI.

[18]  Jürgen Schmidhuber,et al.  Formal Theory of Fun and Creativity , 2010, ECML/PKDD.

[19]  Pascal Hitzler,et al.  Artificial general intelligence: Proceedings of the Second Conference on Artificial General Intelligence, AGI 2009, Arlington, Virginia, USA, March 6-9, 2009 , 2009 .

[20]  Jürgen Schmidhuber,et al.  AUTONOMOUS ACQUISITION OF NATURAL LANGUAGE , 2014, ICIS 2014.

[21]  Jürgen Schmidhuber,et al.  A Family of Gödel Machine Implementations , 2011, AGI.

[22]  The Need for a Revision of Bloom ’ s Taxonomy , .