GA: A Comprehensive Survey on LLM-based GUI Agent
Bibliographic record
Abstract
The Graphical User Interface (GUI) is a visual method that allows users to interact with computers and mobile devices. Nowadays, users rely on GUI for completing some tasks, such as browsing web or using mobile applications. Users often meet some needs such as setting an alarm for 8:00 AM to wake them up and checking the weather for tomorrow. Some commercial agents have been integrated into users personal phones to help the user accomplish a series of basic tasks. Unfortunately, these commercial agents often relied on fixed templates or program scripts to ensure reliability. This also limited their functionality to some basic system applications. Recently large language models (LLMs) have made significant breakthroughs in natural language processing (NLP). Astonishingly, LLMs have demonstrated not only a strong ability to understand and generate text but also planning and reasoning capabilities. Some researchers have considered using LLMs as the agent’s brain, equipping these agents with corresponding capabilities. LLM-based agents are also being applied to help users automate tasks on their personal phones and computers. These agents often can understand the GUI environment on personal phones and computers, allowing them to make decisions to complete tasks. This is also the origin of the term “GUI Agent”. Our review surveys recent research on LLM-based GUI Agents. We summarize the capabilities of existing GUI Agents and also discuss the GUI Agent task automation pipeline. A comprehensive list of studies in this paper will be available at a GitHub repositories.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.004 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.011 | 0.007 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".