On the Legg-Hutter Definition of 'Universal Intelligence'
Presentation by Jobst Landbrebe, (Cognotekt, GmbH)
Department of Computer Science, University at Buffalo, 12 noon, March 12, 2021,
The presentation by Jobst Landgrebe starts with a detailed account of the differences between simple and complex systems and of the reasons why the latter cannot be modelled mathematically in such a way as to allow exact predictions. This implies limits to the degree to which we can use data derived from complex systems to train neural networks in order to create emulations of the behavior of such systems in a machine. This in turn -- since human brains a complex systems -- implies limits also as to the degree to which we can emulate human intelligence in a machine. There follows a critical analysis of the influential Legg-Hutter definition of 'universal intelligence', published in the journal Minds and Machines in 2007.