Biasing AI?
- PDF / 184,653 Bytes
- 4 Pages / 595.224 x 790.955 pts Page_size
- 62 Downloads / 204 Views
REVIEW
Biasing AI? A New Approach to AI Intertwining Natural and Cultural Heuristics Jordi Vallverdu´ 1 Accepted: 22 October 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract AI has achieved great results using bioinspired ideas. Taking into consideration that information is not a fact to be captured but a process that involves both (even minimal) minds and bodies, the author suggests to use a new two-dimensional or blended approach for the design of AI systems: cultural and natural heuristics, including their “biased” mechanisms or implementations. Keywords Heuristics · Bioinspired · Bias · Cultural · Blended cognition · AI
1 Do Not Underestimate Bioinspiration Biological systems have been a constant resource of inspiration for all kind of scientists and engineers. Wonderful sketches of Leonardo show us how bird wings inspired his flying machines, Norbert Wiener revolutionary on Cybernetics were extracted from animal communication analysis, and the obvious McCulloch & Pitts ideas on artificial neural networks were also a way of (re)formalizing artificially the successful evidences provided by natural evolutionary selection. Some of such bioinspired techniques have led to successful research results (genetic algorithms, ANN, machine learning, cellular automata, Alife, Lindenmayer systems, sensor networks, learning classifying systems, emergent systems, membrane computers, neurochips,..), taking most of times the brain and the central nervous system (henceforth, CNS) as fundamental references. Since the polymath John Von Neumann wrote in 1958 his influential book The Computer and the Brain, several generations of AI and cognitive sciences were spellbound by such powerful metaphor: the brain was like a computer, and vice versa. Such engineering approach to the cognitive processes was extremely successful, basically because introduced a mechanistic and simple way to deal with those informational processes, which started to abandon the darkness of Jordi Vallverd´u
[email protected] 1
ICREA Acad`emia - Philosophy Department, UAB, Bellaterra, Barcelona, Spain
centuries captured by the imaginative but opaque minds of philosophers. On the other hand, such approach was as useful for practical purposes as showed to be misleading: human thinking was closer to the heuristic opportunism than we could expect from rational agents, and the emotional role in cognition was not a noisy problem (as previously defended) but instead, it led a fundamental role all across all the process of cognitive steps. Therefore, an opposite approach to artificial cognition has been followed: bioinspired vs. formal. At the very early steps of AI, coming back to the Dartmouth Summer Research Project on Artificial Intelligence in 1956 was centered into formal aspects of artificial reasoning, achieving impressing initial results like Logic Theorist, and later in recent times Deep Blue playing Chess or AlphaGo Zero playing Go, both beating best ever human players (Gary Kasparov and Lee Se-Dol, respectively).
Data Loading...
