Artificial General Intelligence
Professor Danko Nikolic’s thoughts and theories are an inspiration for our work on studying Opentech AI architecture and progress – especially regarding progress towards a more human like general AI. He presents a holistic view on AI and it’s high level organisation in terms of needed system layers in comparison to biological evolution and intelligence. He has also presented a theories in the interface of human and artificial intelligence (ideasthesia) and evolution (practopoiesis). Based on his theories he has presented a view on developing AI systems (AI-Kindergarten), where the intelligence of the AI system is not pre-programmed or pre-trained, but gained via experiences from interaction with the environment of the AI system. A list of scientific publications of Prof. Nikolic recommended as further reading is provided below:
- Nikolić, D. (2017) Why deep neural nets cannot ever match biological intelligence and what to do about it?. International Journal of Automation and Computing, Springer, , Vol. 14, Iss. 5, pp 532–541.
- Nikolić, D. (2015). Only T3-AI can reach human-level intelligence: A variety argument. arXiv arXiv:1505.00775v2.
A distinctive feature of AI systems, in comparison to other types of software intensive systems, is that those have specific structures and processes for cognition & behavior, while perceiving and actuating in their environment. Formalization of these structures and processes is generally known as the cognitive architecture of an AI system – a higly relevant architectural viewpoint for development of AI systems. A comprehensive review article summarizing the progress of past 40 years in research and development of cognitive architectures has been published in 2016:
- Kotseruba, I., Gonzalez, O. J. A., & Tsotsos, J. K. (2016). A Review of 40 Years of Cognitive Architecture Research: Focus on Perception, Attention, Learning and Applications. arXiv preprint arXiv:1610.08602.