One of the biggest difficulties faced by chatbots and other AI packages based on large language models is that they are black box systems which lack transparency. The same difficulty can also apply to large conventionally programmed commercial applications where the lack of transparency can have serious consequences. This problem has been spectacularly demonstrated by the problems caused by the General Post Office's Horizon software where innocent postmasters ended in prison because of errors in the "black box" software. Part of the problem was that the courts were prepared to accept as reliable "evidence" from a complex "black box" system which was unable to answer questions about the origins or reliability of the evidence it was being provided.
CODIL was a very early attempt to build a human-friendly language for a computer which could work with humans as a transparent "electronic clerk" - avoiding the "black box" problems associated with modern large language artificial intelligent systems.. The current study suggested that CODIL worked by modelling how the human brain handles complex information processing tasks. The CODIL archives suggest effective ways of building transparent AI systems and modelling how human intelligence evolved
Showing posts with label ICL. Show all posts
Showing posts with label ICL. Show all posts
Wednesday, January 31, 2024
Subscribe to:
Posts (Atom)