Today, Artificial Intelligence (AI) and automation have become the driving forces behind creativity and human life as a whole. Various industries are now progressively leveraging AI to carry out a wide range of activities, from creative fields to technical sectors like finance and IT.

Some of the most frequently utilized AI platforms include Adobe Firefly, Gemini, and ChatGPT, among others.

By utilizing these tools, anyone can now generate outputs more easily and streamline their workflow. Friends of UNDIRA, this is made possible by crafting commands for the AI to maximize its performance and output—a process commonly known as Prompt Engineering.

Prompt Engineering is the practice of designing specific and contextual instructions for an AI to ensure it delivers the most accurate and desirable output according to the user's intent.

The ability of AI to respond intelligently to commands is rooted in its architecture, which consists of complex neural networks. These networks "learn" from massive-scale datasets, which are continuously enriched by input from users worldwide.

Elaborating on this, a lecturer from Dian Nusantara University (UNDIRA) in the DosenBicara series explained that AI operates optimally due to architectures like Generative Adversarial Networks (GANs). GANs enable AI not only to absorb data as knowledge but also to adapt dynamically. This process is what allows AI to produce outputs that are not only relevant but also increasingly accurate and realistic over time.

Although AI can significantly enhance and shorten our work processes, Friends of UNDIRA, we still sometimes encounter outputs that seem irrelevant or are entirely incorrect. Therefore, here are several tips that can be applied when engaging in Prompt Engineering.

1. Explain Your Objective Thoroughly

In its function as a tool, AI acts as our support system for producing works or data. Therefore, the accuracy and the final result heavily depend on our precision in describing our goals and target demographics.

Provide prompts that are as specific and detailed as possible so the AI can deliver an output with a minimal error rate.

2. Use Technical Instructions (If Necessary)

To achieve the most accurate output possible, apply technical parameters where needed for better and more focused results.

Example: “I am planning to build a gas station franchise in an urban area. Provide me with a calculation for safe distances and safety measures based on the guidelines of ISO 14001 for environmental management and ISO 45001 for occupational health and safety.”

3. Iterate and Improve

As a system that receives data from various sources, AI can have flaws when interpreting results.

As explained by Dr. Pratama Persadha, M.Sos., MM., Chairman of the cybersecurity research institute CISSREC, and Prof. Teddy Mantoro, Professor in Computer Science at Nusa Putra University, during the academic oration at UNDIRA's new student orientation on August 31, 2025, about 90% of the data in AI's memory is "junk."

Therefore, in prompt engineering, we must cross-check and compare different versions of a prompt to minimize the percentage of errors.

4. Use a Clear Structure and Format

In its role as a tool, Prompt Engineering—especially for generating data sheets and code—can sometimes produce outputs that are inconsistent or cannot be properly implemented in applications or data analysis.

This significantly impacts the accuracy and efficiency of a data analyst or programmer, as AI is fundamentally a machine that operates on literal instructions without the ability to guess human intent or intuition.

Therefore, be as specific as possible regarding the desired system structure (such as Python, Ruby, C, or CSS), along with the intended output and its overall function.

The quality of the output—be it code, data analysis, or a creative piece—depends entirely on our skill in designing precise commands through Prompt Engineering. This is not merely a technical skill but an art form that blends logic, creativity, and a deep understanding of how the technology works.

This is where the role of higher education becomes crucial. The Informatics Engineering Program at Dian Nusantara University (UNDIRA) is designed to equip you not only with the ability to use AI tools but also with a comprehensive understanding of AI itself—from data structures and algorithms to neural network architectures. Furthermore, the Informatics Engineering program at UNDIRA is supported by facilities like a Computer Lab, allowing students to get hands-on experience with programming and AI implementation.

If you aspire to be an architect of future technology rather than just a user, we invite you to join us at UNDIRA to turn every instruction into an innovation.

Source of References: 

Prompt Engineering 101: Panduan Lengkap untuk Hasil Optimal

Dosen Bicara Teknik Informatika - AI mampu menggantikan programmer? Begini cara kerja AI