Challenges to Using Large Language Models in Code Generation and Repair

Liliana Pasquale; Antonino Sabetta; Marcelo d'Amorim; Peter Hegedus; Mehdi Tarrit Mirakhorli; Hamed Okhravi; Mathias Payer; Awais Rashid; Joanna C.S. Santos; Jonathan M. Spring; Lin Tan; Katja Tuma; Fabio Massacci

doi:10.1109/MSEC.2025.3530488

Challenges to Using Large Language Models in Code Generation and Repair

Liliana Pasquale
, Antonino Sabetta^*
, Marcelo d'Amorim
, Peter Hegedus
, Mehdi Tarrit Mirakhorli
, Hamed Okhravi
, Mathias Payer
, Awais Rashid
, Joanna C.S. Santos
, Jonathan M. Spring
, Lin Tan
, Katja Tuma
, Fabio Massacci (Editor)

^*Corresponding author for this work

Research output: Contribution to Journal › Article › Academic

150 Downloads (Pure)

Abstract

Large language models (LLMs) hold great promise in solving many challenges arising from software complexity, including the possibility of automating code generation and repair. Although we cannot deny the groundbreaking nature of LLM-based code repair, we must be realistic in positioning current results.

Original language	English
Pages (from-to)	81-88
Number of pages	8
Journal	IEEE Security and Privacy
Volume	23
Issue number	2
Early online date	26 Mar 2025
DOIs	https://doi.org/10.1109/MSEC.2025.3530488
Publication status	Published - Apr 2025

Bibliographical note

Publisher Copyright:
© 2003-2012 IEEE.

Access to Document

10.1109/MSEC.2025.3530488

Challenges_to_Using_Large_Language_Models_in_Code_Generation_and_RepairFinal published version, 3.54 MBLicence: Article 25fa Dutch Copyright Act

Persistent URL (handle)

Persistent link

Cite this

@article{28418e18f3224e2eb64a8c657adc84e5,

title = "Challenges to Using Large Language Models in Code Generation and Repair",

abstract = "Large language models (LLMs) hold great promise in solving many challenges arising from software complexity, including the possibility of automating code generation and repair. Although we cannot deny the groundbreaking nature of LLM-based code repair, we must be realistic in positioning current results.",

author = "Liliana Pasquale and Antonino Sabetta and Marcelo d'Amorim and Peter Hegedus and Mirakhorli, \{Mehdi Tarrit\} and Hamed Okhravi and Mathias Payer and Awais Rashid and Santos, \{Joanna C.S.\} and Spring, \{Jonathan M.\} and Lin Tan and Katja Tuma and Fabio Massacci",

note = "Publisher Copyright: {\textcopyright} 2003-2012 IEEE.",

year = "2025",

month = apr,

doi = "10.1109/MSEC.2025.3530488",

language = "English",

volume = "23",

pages = "81--88",

journal = "IEEE Security and Privacy",

issn = "1540-7993",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "2",

}

TY - JOUR

T1 - Challenges to Using Large Language Models in Code Generation and Repair

AU - Pasquale, Liliana

AU - Sabetta, Antonino

AU - d'Amorim, Marcelo

AU - Hegedus, Peter

AU - Mirakhorli, Mehdi Tarrit

AU - Okhravi, Hamed

AU - Payer, Mathias

AU - Rashid, Awais

AU - Santos, Joanna C.S.

AU - Spring, Jonathan M.

AU - Tan, Lin

AU - Tuma, Katja

A2 - Massacci, Fabio

PY - 2025/4

Y1 - 2025/4

N2 - Large language models (LLMs) hold great promise in solving many challenges arising from software complexity, including the possibility of automating code generation and repair. Although we cannot deny the groundbreaking nature of LLM-based code repair, we must be realistic in positioning current results.

AB - Large language models (LLMs) hold great promise in solving many challenges arising from software complexity, including the possibility of automating code generation and repair. Although we cannot deny the groundbreaking nature of LLM-based code repair, we must be realistic in positioning current results.

UR - https://www.scopus.com/pages/publications/105004222524

UR - https://www.scopus.com/pages/publications/105004222524#tab=citedBy

U2 - 10.1109/MSEC.2025.3530488

DO - 10.1109/MSEC.2025.3530488

M3 - Article

AN - SCOPUS:105004222524

SN - 1540-7993

VL - 23

SP - 81

EP - 88

JO - IEEE Security and Privacy

JF - IEEE Security and Privacy

IS - 2

ER -

Challenges to Using Large Language Models in Code Generation and Repair

Abstract

Bibliographical note

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this