Speaker
Description
The use of Artificial Intelligence (AI) has become an inseparable part of the software development process, parts of which are routine, requiring precision rather than creativity. AI code generation speeds up productivity by automatically completing code fragments and frees up time and brainpower to perform more tasks that require creativity or problem-solving. When writing code, the programmer uses not only the accumulated experience, but often also intuition, which helps him assess both the correctness of the code and the implications of its use.
At the same time, AI uses different and increasingly sophisticated models and a thought chain resembling human thinking through modern approaches to Large Language Models (LLM) and Neural Networks. However, increasingly serious problems are emerging when using AI.
The publication proposes a model of an AI algorithm with two functionally different cores: one creative and one process, including control functions. This model overcomes a large part of the identified shortcomings of AI and complements their development. The model is supplemented with quantitative metrics to determine the economic efficiency of its application.