Tasks for aligning human and machine planning

Publication date: October 2019Source: Current Opinion in Behavioral Sciences, Volume 29Author(s): Bas van Opheusden, Wei Ji MaResearch on artificial intelligence and research on human intelligence rely on similar conceptual foundations and have long inspired each other [1,2•]. However, achieving concrete synergy has been difficult, with one obstacle being a lack of alignment of the tasks used in both fields. Artificial intelligence research has traditionally focused on tasks that are challenging to solve, often using human performance as a benchmark to surpass [3, 4, 5, 6, 7]. By contrast, cognitive science and psychology have moved towards tasks that are simple enough to allow for detailed computational modeling of people’s choices. These divergent objectives have led to a divide in the complexity of tasks studied, both in perception and cognition. The purpose of this paper is to explore the middle ground: are there tasks that are reasonably attractive to both fields and could provide fertile ground for synergy?

Source: Current Opinion in Behavioral Sciences - August 30, 2019 Category: Psychiatry & Psychology Source Type: research