ROUND TABLE: About the limitations of artificial intelligence
Natal'ya V. Dovgalenko Yuri Gagarin State Technical University of Saratov, Saratov, Russia
Abstract
The report presents the problem of limitations of artificial intelligence, which the author examines through two methodological programs. The first program was put forward in the XXth century and presented the formal-logical and mathematical impossibility of AI competition with human intelligence. R. Penrose expressed her ideas most fully: intelligence cannot solve a problem while remaining within a closed system of axioms. Getting out of it requires «understanding», that AI cannot achieve. The second program, proposed in the XXIst century, relies on a neural network approach and achieves incredible success by simulating natural intelligence and neural activity. In this program, limitations are associated with the specific features of the data (neural connections) that are being simulated, the volume of data (memory growth), and the control of growth.
Speaker
Natal'ya V. Dovgalenko
Yuri Gagarin State Technical University of Saratov, Department of History and Philosophy
Russia
Discussion
Ask question
