Technical Debt (TD) is the implied cost of additional rework caused by choosing easier solutions in favour of shorter release time. It impacts software maintainability and evolvability, manifesting as different types (e.g., Code, Test, Architecture). Algorithm Debt (AD) is a new TD type recently identified as sub-optimal implementations of algorithm logic in scientific and Artificial Intelligence (AI) software. Given its newness, AD and its impact on AI-driven software remains a research gap. This poster aims to motivate reflective discussion on AD in AI software, by summarising findings, discussing its possible impact, and outlining future areas of work.
@inproceedings{Simon2023,
author = {Simon, Iko-ojo and Vidoni, Melina and Fard, Fatemeh},
title = "{Algorithm Debt: Challenges and Future Paths}",
year = {2023},
isbn = {},
publisher = {Association for Computing Machinery},
address = {New York, USA},
doi = {},
booktitle = "{2nd International Conference on AI Engineering – Software Engineering for AI (CAIN'23)}",
pages = {90-91},
numpages = {12},
location = {Melbourne, Australia},
series = {CAIN'23}
}