On 26-27th September 2022 the Royal Society, London, held a discussion meeting on "Cognitive Artificial Intelligence" at which the following poster was displayed:
A Possible Model of the Brain's Symbolic Assembly Language?
CODIL
An early “Blue Sky” research idea that was “abandoned” because it was incompatible with Good Old-Fashioned Artificial Intelligence
§The CODIL research was triggered by a 1967 study of the salesman/clerical interface of a large 1970s sales accounting system.
§ A key design factor was transparency as an essential requirement was that the computer must always be able to explain what it was doing in human terms.
§ CODIL is a task independent language framework, with a very simple syntax, which described tasks in terms of sets and set operations, rather than “program” and “data.”
§ It maps onto a recursive network, where every node can represent a set, a simple subset of a set, or the description of a set.
§ The “current context” models human short-term memory.
§ The language was powerful enough to support a powerful heuristic
problem solver as well as database and educational applications.
§ BUT – People who have leant a procedural programming language
find the CODIL approach counter-intuitive.
So what next? Should the research restart? And if so by whom and where? Because of my age the research records could end up in a skip if no-one is interested?
No comments:
Post a Comment