The article discusses a proposal of using artificial intelligence (AI) methods as a tool for public debt management optimisation. This proposal is a continuation of some earlier work by the authors. The following AI methods suggested for this purpose are described: artificial neural networks, theory of fuzzy sets, genetic algorithms, methods of knowledge representation. The authors also discuss potential fields for their application in public debt management. The management is understood here in a very broad sense and covers analyses, diagnoses and forecasts preparation as well as decision making process. The proposal presented should be treated as a starting point for further research which requires participation of scientists specialised in artificial intelligence. Since public debt constitutes a serious problem for many countries, it seems reasonable that a project for the implementation of AI methods in public debt management should be an outcome of international cooperation, for instance under the aegis of EU, OECD or IMF, with reliance on Polish scientists.