HiLDe: Intentional Code Generation via Human-in-the-Loop Decoding
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918041017647104 |
|---|---|
| author | González, Emmanuel Anaya Rothkopf, Raven Lerner, Sorin Polikarpova, Nadia |
| author_facet | González, Emmanuel Anaya Rothkopf, Raven Lerner, Sorin Polikarpova, Nadia |
| contents | While AI programming tools hold the promise of increasing programmers' capabilities and productivity to a remarkable degree, they often exclude users from essential decision-making processes, causing many to effectively "turn off their brains" and over-rely on solutions provided by these systems. These behaviors can have severe consequences in critical domains, like software security. We propose Human-in-the-loop Decoding, a novel interaction technique that allows users to observe and directly influence LLM decisions during code generation, in order to align the model's output with their personal requirements. We implement this technique in HiLDe, a code completion assistant that highlights critical decisions made by the LLM and provides local alternatives for the user to explore. In a within-subjects study (N=18) on security-related tasks, we found that HiLDe led participants to generate significantly fewer vulnerabilities and better align code generation with their goals compared to a traditional code completion assistant. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_22906 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | HiLDe: Intentional Code Generation via Human-in-the-Loop Decoding González, Emmanuel Anaya Rothkopf, Raven Lerner, Sorin Polikarpova, Nadia Human-Computer Interaction Artificial Intelligence Programming Languages While AI programming tools hold the promise of increasing programmers' capabilities and productivity to a remarkable degree, they often exclude users from essential decision-making processes, causing many to effectively "turn off their brains" and over-rely on solutions provided by these systems. These behaviors can have severe consequences in critical domains, like software security. We propose Human-in-the-loop Decoding, a novel interaction technique that allows users to observe and directly influence LLM decisions during code generation, in order to align the model's output with their personal requirements. We implement this technique in HiLDe, a code completion assistant that highlights critical decisions made by the LLM and provides local alternatives for the user to explore. In a within-subjects study (N=18) on security-related tasks, we found that HiLDe led participants to generate significantly fewer vulnerabilities and better align code generation with their goals compared to a traditional code completion assistant. |
| title | HiLDe: Intentional Code Generation via Human-in-the-Loop Decoding |
| topic | Human-Computer Interaction Artificial Intelligence Programming Languages |
| url | https://arxiv.org/abs/2505.22906 |