The Information and Communications Technology is Driving Artificial Intelligence to Leverage Refined Knowledge for the World Sustainable Development -Part 3-

The paper approaches the analysis of the complex processes of artificial intelligence contribution to knowledge refining, in the general context of Information and Communications Technology (ICT) exponential evolution as main driving factor of the progress of the Information society (IS) toward Knowledge Based Society (KBS). In this context ICT grows, but considering its complex proliferation and exponential development at Earth scale, lately by artificial intelligence (AI), as leading to a planetary digital disruption, it is necessary to estimate how big would be this iceberg, because this growing with no precisely known perspective offers enough reasons to be timely and responsibly analysed. The analysis is a dynamic challenge because the question has multiple angles to be answered, mainly twofold: how much ICT can really grow in its diverse directions (applications fields) and, on the other side, to which extent the specific and then global consequences of these growings (advances) are sustainable for humankind and Earth? The author presented, by a systemic approach, some relevant examples of this complex space/time plan where ICT/AI is fast evolving, aiming both the growing and the knowledge refining potentials. The domain of understanding natural language (UNL) was selected first just because it is actually one of the most prominent, prolific and challenging fields of AI, without mentioning the essential role of man-machine communication in this digital poque of humankind evolution and generally in the IS/KBS processes. UNL rises difficult problems of accuracy (false understanding) in the important applications for security, where the performance is critical for avoiding false identification or service denial. As the impact of AI to the health care field is enough important for humankind, it is developed and analysed with priority and perhaps one of the most prominent work in the AI use for the medical field is done by IBM, but the detailed analysis of this case showed that IBM powerful technology (Watson) is not enough for the complex reality of today s health care system. At least when trying to apply Watson to cancer treatment, one of medicine s biggest challenges, IBM faced a big difference between the way machines learn and the way doctors work. One of the conclusions is that in the ICT/AI struggle to win higher and higher peaks of performance, especially in the health care field, where humankind expectations are naturally the most justified, all the progress efforts must be supported, encouraged but also carefully and realistically timely analysed in order to obtain the optimal results, even when the downs appear to be prominent. It is also very important to use and learn lessons from all the refined knowledge these ups and downs of ICT/AI provide, adding new human expertise and use all available resources to get sustainable progress in all activity areas, keeping humankind health as first priority, not only in the ICT/AI direct health care applications, but also in the indirect consequences that could come sooner or later from other fields progress. The paper analysis includes another relevant systemic example of AI growing, the communications vast area, because, albeit the idea of intelligent network is not new, the actual AI potential and the trend of using most of wireless communication capacity are generating very important applications aiming optimization. Considering the complex picture of the factors that could influence wireless communications performance, it seems that the medium access control is a zone where they could benefit most from AI(machine learning) support in optimizing areas like Signal detection, Channel encoding and decoding, Channel estimation, prediction, and compression, Resource allocation etc. An important conclusion of paper is that extending applications area of AI could be less efficient if we do not add some new approaches of the ways to understand and use the complex relations between data, information and knowledge resulting from these applications and more than these, add the cognitive principles we must develop AI on, in order to perform towards human brain-like models that leverage learning from that knowledge, i.e. properly refining knowledge. Another conclusion is the necessity of conceiving and using the new systems/technologies with a balance between perpetual human values/principles and the rational actual trends. There is still a reasonable expectation that ICT/AI could better leverage the management of this balance by their implementation in most advanced cognitive systems, i.e. we have to keep alive wisdom principles and update them with the refined knowledge that reflects the actual needs of humankind and Earth, not only for immediate efficiency, but for a stable future of the available resources and life environment. The actual challenges of ICT/AI development must be synchronized with the humankind surviving (on Earth) challenges and risks, but here the point is that, more than AI, the cognitive computing systems (CCS) could really be a part of the right decision in a matter that is closing now to the human intelligence stage, just by the potential to approach high level problems. Considering the high level of cognitive science, in the concrete development of AI toward CCS there is a long way, naturally similar to the IBM-Watson bitter-sweet evolution, i.e. very difficult and complex. Going to the core of CCS, the aim is the human brain map and functionality, which must be replicated in the future. The paper also referred to concrete principles, methods and technologies to reach such performance and complex objectives, which are among the most sophisticated ingredients of the evolution from AI to CCS, without neglecting brain complex operating processes and the ways these are related to reality. For a progress from here ahead it is very probable to refer to the people s changing behaviour and evolutionary epistemology, or simple to the brain. Without offering solutions here, this philosophy approach (one of the many possible!) has the quality to reveal the complexity of the AI target to follow the human brain model and in the same time is a good instrument of orienting the AI/CCS design to overpass inherent human limits of reflecting reality, by using objective measures/metrics for sensing most of reality parts, but unfortunately not for all. Another useful conclusion is that the most difficult problems for AI/CCS progress are in the aria of managing knowledge associated to abstract concepts (like compassion, causality or democracy) where never the human perception, imagination and creation would be completely replaced. Consequently AI/CCS should be developed in a close collaboration between AI and human intelligence, expertise and responsibility, i.e. CCS should include selected humans in all processes, from design to implementation and usage, as intrinsic parts of the systems and applications, valuing the AI natural language processing, computer vision or other similar advanced man-machine interfaces (like emerging brain-computer interfaces). Although all the above (at least) human contributions are important, it is worth to notice that the most prominent should be in responsibly designing AI/CCS as the knowledge used and refined must have the potential to generate sustainable progress of humankind and Earth ecosystem.