Author(s)

Nikolaos Panagiotopoulos

On Board Computers & Data Handling Systems, European Space Research & Technology Centre (ESTEC), Noordwijk, Netherlands.

Although Artificial Intelligence (AI) and Machine Learning (ML) are attracting a lot of scientific and engineering attention nowadays, nothing up to now has been achieved to reach the level of building machines that possess human-like intelligence. Though, the engineering community continuously claims that several engineering problems are solved using AI or ML. Here, it is argued that engineers are not being able to build intelligent machines, implying that the systems claimed to have AI/ML belong to different engineering domains. The base of the syllogism is the existence of four main obstacles on which extensive elucidation is performed. In addition, attempt to clear out the developed confusion, mis-use and ab-use of the phrases “Artificial Intelligence” and “Machine Learning” by scientists and engineers is carried out. Furthermore, mathematical, and philosophical approaches are also mentioned that strengthen the argument against AI implementability as part of the whole syllogism. Finally, an alternative approach (not being unique) is suggested and discussed for performing research on AI and ML by the engineers. It is based on complexity theory and non-linear adaptive systems and provides the benefit of eliminating the before mentioned pragmatic and philosophical obstacles that engineers are facing and ignoring, without creating confusion on this scientific endeavor.

Artificial Intelligence, Machine Learning, Complexity, Definitions, Requirements, Verification and Validation, Logic, Emergence Properties

Panagiotopoulos, N. (2023) Can Engineers Build AI/ML Systems? Analysis and an Alternative Approach. Open Journal of Philosophy, 13, 504-530. doi: 10.4236/ojpp.2023.133034.