In addressing the need for test case generation in software projects and the validation and repair processes, various algorithms and AI models are increasingly being applied with novel approaches. On the other hand, despite the established effectiveness of the Test-Driven Development (TDD) approach in testing and development, there is still a lack of research examining the impact of human-machine interaction on software validation and coding. This paper introduces a tool, the test-skeleton generator, which utilizes an OpenAI model to generate test skeletons. These skeletons include test names, signatures, and scenario descriptions, omitting the actual test bodies. To explore the implications of this tool, an empirical experiment involving student participation was conducted to assess the conversion of test skeletons into functional tests with human-machine interaction. The study reveals significant insights, indicating that human- machine interaction plays a crucial role in shaping both the test- ing and programming phases, encouraging students to prioritize writing tests before modifying source code. Teams adopting this approach demonstrate a tendency to produce more tests, leading to higher code coverage. Additionally, our research underscores the growing potential of AI language models to generate tests that closely resemble those written by human developers. Notably, human-machine interaction has proven its significant positive impact on the validation and repair process of AI-generated tests.