Some reasons for artificial intelligence errors in solving combinatorial problems
Abstract:
The paper considers some relatively simple exercises in combinatorics that artificial intelligence solves with errors. The ChartGPT and DeepSeek systems are considered as representatives of artificial intelligence. Screenshots with questions (prompts) to these systems and their answers are provided. Hypotheses are put forward about the causes of artificial intelligence errors in solving the problems under consideration. It is assumed that such errors can occur when using artificial intelligence for software development. The topics of further research are put forward that may be interesting for determining the conditions for further use of artificial intelligence.
