ARC-AGI 1, 2 explainable and sustainable solution via interaction space theory

Description

We propose an interdisciplinary project between computer science and mathematics in explainable and sustainable artificial intelligence via Interaction Space theory under the supervision of Paolo Giordano. The candidate should be interested in the following fields: 1) Explainable and sustainable artificial intelligence 2) Mathematical modelling of complex systems and their applications 3) Critical approaches to generative AI. The primary objective of this project is the development of a novel approach to explainable and sustainable artificial intelligence (XSAI) based on interaction space (IS) theory (IST), a novel unifying mathematical theory of complex systems (CS) which has the property to include cellular automata, agent based models, master equation based models, continuous or discrete dynamical systems, networked dynamical models, artificial neural networks and genetic algorithms in a single mathematical notion. Our AI approach aims to make sustainable AI accessible to all countries and not solely to rich ones or big tech companies. A major goal is to accelerate a meaningful development of AI based on a human inspired notion of cause-effect (CE) truth (CE based world model). The student has to contribute on the present development of the Python code for a solution in natural language of ARC-AGI 1, 2 benchmarks, running in a normal laptop and without using an internet connection, a solution which is out of the scope of present approaches. The candidate should have a good knowledge of computer programming and experience in mathematical modelling. A preliminary knowledge of IST is not required and can be fulfilled during the first part of the research work. To get an idea of the research area and the required skills, please have a look at the following web sites and related bibliographical references: • https://arcprize.org/arc-agi • http://www.mat.univie.ac.at/~giordap7/ • https://sites.google.com/ugm.ac.id/xaiugm/ • http://www.mat.univie.ac.at/~giordap7/XSAI.pdf • https://www.mat.univie.ac.at/~giordap7/RLE.mp4 • https://www.mat.univie.ac.at/~giordap7/RLS.mp4 • https://github.com/giopao/ARC-AGI1