Introduction: Artificial intelligence should meet natural stupidity. But it cannot!
Riccardo Viale · Edward Elgar Publishing eBooks · 2023
The simulation of human mind might be possible with the advancement of cognitive science research on memory, thinking, reasoning, emotion, decision making and embodied cognition. But the attempt should avoid the Rationality Bias of “Human Intelligence as Optimal Computation Device”, that is thinking that the mind is mainly the ability to process optimally data and to conform to the principle of logic, probability and utility. On the contrary AI should consider the imperfect formal features of human intelligence and how many times it is adaptively better to be biased than to be formally perfect. The chapter presents some theoretical and empirical reasons in support of this argument. The two main theses that will be developed are 1) The current programs of emulative AI, based on the processing of Big Data through machine and deep learning programs, are flawed in many contexts and in particular they manage to find patterns of past data, but they are not able to forecast future unstable events, such as financial ones, characterized by complexity and uncertainty; 2) The Simulative AI, also called Psychology AI (Gigerenzer, 2022), despite being more suited to grasping more realistic aspects of human intelligence, in fact has not been able to achieve this goal due to the lack of the body dimension of the simulated cognitive experience . In any case, according to some philosophers, the construction of machines that have the same intelligence as man is an a priori impossible undertaking.