Prompt Strategy for Optimal AI Results

Abstract

The rapid development of artificial intelligence (AI) technology requires users to have the ability to compose effective instructions or prompts so that the output produced meets their needs. However, many users, especially students, do not yet understand prompting strategies optimally. Therefore, this community service activity was carried out in the form of a webinar and workshop entitled “Prompt Strategies for Optimal AI Results.” This activity aims to improve participants' understanding and skills in utilizing AI through prompting techniques such as Chain-of-Thought (CoT), Few-shot Prompting, Retrieval-Augmented Generation (RAG), ReAct, and Tree of Thoughts (ToT). The methods used in this activity included socialization, pre-testing, webinar material delivery, hands-on practice in the workshop session, post-testing, and evaluation through a feedback questionnaire. The evaluation results showed a significant increase in participants' understanding, marked by an increase in the average score from 78.06 in the pre-test to 100 in the post-test, or an increase of 28.1%. Additionally, participant satisfaction with the material, presenters, and activity implementation was very high. Thus, this activity proved effective in improving participants' literacy and competence in the optimal use of AI.