ANALISIS PENGGUNAAN LARGE LANGUAGE MODEL (LLM) DALAM PROSES BELAJAR MAHASISWA

Hasan, Miranda A. (2026) ANALISIS PENGGUNAAN LARGE LANGUAGE MODEL (LLM) DALAM PROSES BELAJAR MAHASISWA. Sarjana thesis, Universitas Muhammadiyah Gorontalo.

[img] Text (Abstrak Miranda)
ABSTRACT.pdf

Download (156kB)
[img] Text (Bab 1 Miranda Hasan)
BAB I.pdf

Download (123kB)
[img] Text (Dapus Miranda Hasan)
DAFTAR PUSTAKA.pdf

Download (123kB)
[img] Text (Skripsi Miranda Hasan)
SKRIPSI MIRANDA A. HASAN.pdf
Restricted to Repository staff only

Download (4MB)

Abstract

The transformation of artificial intelligence through Large Language Models (LLMs) such as ChatGPT, Gemini, and Claude has shifted the paradigm of information retrieval and academic task completion among students. However, this convenience raises concerns regarding the decline of critical thinking skills and the risk of students falling into The Hallucination Trap. This study aims to map the patterns of LLM usage, evaluate the efficiency level of the learning process, and measure the impact of students' technical competence in prompt engineering on information validity at Universitas Muhammadiyah Gorontalo. This type of research is quantitative with a descriptive approach. The sample consisted of 96 active students from various class years selected using the Proportionate Stratified Random Sampling technique. Data analysis was conducted descriptively along with the calculation of the Respondent Achievement Level (TCR) using SPSS software. The results showed that LLMs have become a part of students' routine academic activities, where the most dominant platform used is ChatGPT (38.5%) and the most frequent type of academic task assisted is searching for references (34.5%). From the aspect of learning efficiency, the time efficiency indicator obtained the highest TCR of 88% (Good/Efficient Category), while the verification burden indicator obtained the lowest score of 74% (Sufficient Category). In the prompt engineering and hallucination risk variables, the output validation aspect achieved the highest TCR of 78.48%, while the awareness of hallucination risk stood at 73.89%. It can be concluded that the use of LLMs is considered effective in accelerating student learning productivity, yet its application still requires strict critical thinking and independent verification to ensure that the learning outcomes obtained remain valid and accurate.

Item Type: Thesis (Sarjana)
Contributors:
ContributionContributorsNIDN/NBMEmail
Thesis advisorLasarudin, Alter0923038503UNSPECIFIED
Thesis advisorTupamahu, Frangky0920068702UNSPECIFIED
Additional Information: Transformasi kecerdasan buatan melalui Large Language Model (LLM) seperti ChatGPT, Gemini, dan Claude telah mengubah paradigma pencarian informasi dan pengerjaan tugas akademik mahasiswa. Namun, kemudahan ini memicu kekhawatiran terkait penurunan kemampuan berpikir kritis dan risiko mahasiswa terjebak dalam The Hallucination trap. Penelitian ini bertujuan untuk memetakan pola penggunaan LLM, mengevaluasi tingkat efisiensi proses belajar, serta mengukur dampak kompetensi teknis prompt engineering mahasiswa terhadap validitas informasi di Universitas Muhammadiyah Gorontalo. Jenis penelitian adalah kuantitatif dengan pendekatan deskriptif. Jumlah sampel sebanyak 96 mahasiswa aktif dari berbagai tahun angkatan yang dipilih menggunakan teknik Proportionate Stratified Random Sampling. Analisis data dilakukan secara deskriptif dan perhitungan Tingkat Capaian Responden (TCR) menggunakan bantuan software SPSS. Hasil penelitian menunjukkan bahwa LLM telah menjadi bagian dari aktivitas akademik rutin mahasiswa, di mana platform yang paling dominan digunakan adalah ChatGPT (38,5%) dengan jenis tugas terbanyak adalah mencari referensi (34,5%). Dari aspek efisiensi belajar, indikator efisiensi waktu memperoleh TCR sebesar 78% (Kategori Baik/Efisien), sementara indikator beban verifikasi memperoleh skor sebesar 81% (Kategori Baik/Efisien). Pada variabel prompt engineering dan risiko halusinasi, aspek validasi output meraih TCR tertinggi sebesar 78,48%, sedangkan kesadaran risiko halusinasi berada pada angka 73,89%. Dapat disimpulkan bahwa penggunaan LLM dinilai efektif dalam mengakselerasi produktivitas belajar mahasiswa, namun penerapannya tetap membutuhkan sikap kritis dan verifikasi mandiri yang ketat agar hasil belajar yang diperoleh tetap valid dan akurat.
Uncontrolled Keywords: Large Language Model (LLM), Efisiensi Belajar, UTAUT, Prompt engineering, Hallucination trap
Subjects: T Technology > T Technology (General)
T Technology > TA Engineering (General). Civil engineering (General)
Divisions: Fakultas Sains dan Ilmu Komputer > Program Studi Sistem informasi
Depositing User: Miranda A. Hasan
Date Deposited: 03 Jul 2026 06:38
Last Modified: 03 Jul 2026 06:38
URI: http://repository.umgo.ac.id/id/eprint/7245

Actions (login required)

View Item View Item