KIAM Main page Web Library  •  Publication Searh   
Publication

Conference material: "Proceedings of the 6th International Conference Futurity designing. Digital reality problems (February 2-3, 2023, Moscow)"
Authors: Zhuravlev D.V., Smolin V.S.
The neural network revolution of artificial intelligence and it's development options
Abstract:
The civilization development, which ensures the science and technology progress, has led to the theories of algorithms and control creation and the achievement of technological capabilities to create 'smart' machines, primarily based on information and computing technologies. But, just as in chemistry until the first half of the 19th century there was no doubt that life force was needed to obtain organic substances, so it is now widely believed that all modern successes in artificial intelligence are only an imitation of mental activity, which 'for real' can occur only in a living person brain. To substantiate this opinion, it is indicated that the created networks of formal neurons not only have work algorithms that differ markedly from the living neural networks functioning, but also their learning processes properties are also radically different. Yes, this is true, but modern successes in the development of 'narrow' artificial intelligence provide a basis for understanding the existing unresolved problems and allow us to look for ways to overcome them. The main step is to abandon the imitation of problem solving methods based on the knowledge that a person has and the transition to the acquisition and accumulation of new knowledge through interaction with the environment. But this does not mean that the civilizational knowledge accumulated by mankind should be discarded. On the contrary, it is the existing knowledge that should serve as the basis for obtaining new ones.
Keywords:
formal neurons, strong artificial general intelligence (AGI), civilization, competition, decomposition, knowledge
Publication language: russian,  pages: 22 (p. 223-244)
Russian source text:
Export link to publication in format:   RIS    BibTeX
About authors:
  • Zhuravlev Dmitry Vladimirovich,  ,  CIFROMED LLC
  • Smolin Vladimir Sergeevich,  orcid.org/0000-0001-9030-6545KIAM RAS