Addressing Safety Concerns in AI Systems: An Analysis and Remediation Strategies |
||
 |
 |
|
| © 2024 by IJCTT Journal | ||
| Volume-72 Issue-4 |
||
| Year of Publication : 2024 | ||
| Authors : Ramakrishnan Neelakandan, Vidhya Sankaran | ||
| DOI : 10.14445/22312803/IJCTT-V72I4P107 | ||
How to Cite?
Ramakrishnan Neelakandan, Vidhya Sankaran, "Addressing Safety Concerns in AI Systems: An Analysis and Remediation Strategies ," International Journal of Computer Trends and Technology, vol. 72, no. 4, pp. 58-62, 2024. Crossref, https://doi.org/10.14445/22312803/IJCTT-V72I4P107
Abstract
As Artificial Intelligence (AI) systems become increasingly ubiquitous across various domains, ensuring their safety and trustworthiness is of paramount importance. This paper conducts a comprehensive analysis of the key safety issues associated with AI systems, including lack of transparency and interpretability, data quality and bias issues, vulnerability to adversarial attacks, and ethical considerations. The lack of transparency in many AI models, particularly deep learning systems, makes it challenging to understand their decision-making processes, detect biases or errors, and foster trust among stakeholders. Additionally, AI systems rely on the quality and representativeness of their training data; otherwise, they risk propagating existing biases or introducing new ones, leading to unfair outcomes. Moreover, AI systems have been shown to be vulnerable to adversarial attacks and data poisoning, posing severe risks in safety-critical applications like autonomous vehicles or medical devices. Furthermore, the deployment of AI systems raises ethical concerns regarding their alignment with human values, potential for unintended consequences, and impact on marginalized communities.
To mitigate these safety issues, the paper proposes several remediation strategies, including techniques for enhancing transparency and interpretability, improving data quality and mitigating bias, implementing robust security measures, and adopting human-centered design approaches that prioritize ethical considerations. The paper emphasizes the importance of a multidisciplinary approach involving experts from various fields, establishing clear governance structures, and fostering collaboration among stakeholders to ensure the responsible development and deployment of AI systems.
Keywords
Artificial Intelligence, Safety, Ethical AI, Bias, Data Quality.
Reference
[1] Joy Buolamwini, and Timnit Gebru “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification,” Proceedings of the 1st Conference on Fairness, Accountability and Transparency, vol. 81, pp. 77-91, 2018.
[Google Scholar] [Publisher Link]
[2] Jeffrey Dastin, Amazon Scraps Secret AI Recruiting Tool that Showed Bias against Women, 1st ed., Ethics of Data and Analytics, pp. 1- 4, 2022.
[Google Scholar] [Publisher Link]
[3] Sam Levin, and Julia Carrie Wong, “Self-Driving Uber Kills Arizona Woman in First Fatal Crash Involving Pedestrian,” The Guardian, 2018.
[Google Scholar] [Publisher Link]
[4] Filip Karlo Došilović, Mario Brčić, and Nikica Hlupić, “Explainable Artificial Intelligence: A Survey,” 2018 41st International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), Opatija, Croatia, pp. 210-215, 2018.
[CrossRef] [Google Scholar] [Publisher Link]
[5] Julia Angwin et al., Machine Bias, 1st ed., Ethics of Data and Analytics, pp. 1-11, 2022.
[Google Scholar] [Publisher Link]
[6] Tolga Bolukbasi et al., “Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings,” Advances in Neural Information Processing Systems, pp. 1-9, 2016.
[Google Scholar] [Publisher Link]
[7] C. Garvie, “The Perpetual Line-Up: Unregulated Police Face Recognition in America,” Georgetown Law Center on Privacy & Technology, 2016.
[Google Scholar] [Publisher Link]
[8] Naveed Akhtar, and Ajmal Mian, “Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey,” IEEE Access, vol. 6, pp. 14410-14430, 2018.
[CrossRef] [Google Scholar] [Publisher Link]
[9] Kevin Eykholt et al., “Robust Physical-World Attacks on Deep Learning Visual Classification,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1625-1634, 2018.
[Google Scholar] [Publisher Link]
[10] Ali Shafahi et al., “Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks,” Advances in Neural Information Processing Systems, pp. 1-11, 2018.
[Google Scholar] [Publisher Link]
[11] Ninareh Mehrabi et al., “A Survey on Bias and Fairness in Machine Learning,” ACM Computing Surveys (CSUR), vol. 54, no. 6, pp. 1- 35, 2021.
[CrossRef] [Google Scholar] [Publisher Link]
[12] David Gunning, and David W. Aha, “DARPA’s Explainable Artificial Intelligence (XAI) Program,” AI Magazine, vol. 40, no. 2, pp. 44- 58, 2019.
[CrossRef] [Google Scholar] [Publisher Link]
[13] Geoffrey Hinton, Oriol Vinyals, and Jeff Dean, “Distilling the Knowledge in a Neural Network,” Arxiv, pp. 1-9, 2015.
[CrossRef] [Google Scholar] [Publisher Link]
[14] Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin, “Why Should I Trust You?": Explaining the Predictions of Any Classifier,” Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1135-1144, 2016.
[CrossRef] [Google Scholar] [Publisher Link]
[15] Pedro Saleiro et al., “Aequitas: A Bias and Fairness Audit Toolkit, Arxiv, pp. 1-19, 2018.
[CrossRef] [Google Scholar] [Publisher Link]
[16] Brian Hu Zhang, Blake Lemoine, and Margaret Mitchell, “Mitigating Unwanted Biases with Adversarial Learning,” Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society, pp. 335-340, 2018.
[CrossRef] [Google Scholar] [Publisher Link]
[17] Joy Buolamwini, Artificial Intelligence Has a Problem With Gender and Racial Bias. Here’s How to Solve it, 2019. [Online]. Available: https://time.com/5520558/artificial-intelligence-racial-gender-bias/
[18] Aleksander Madry et al., “Towards Deep Learning Models Resistant to Adversarial Attacks,” Arxiv, pp. 1-28, 2017.
[CrossRef] [Google Scholar] [Publisher Link]
[19] K. Kokalitcheva, How Tesla is Building a Secret Weapon for Security Challenges in Autonomous Driving, 2019. [Online]. Available: https://www.Forbes.com/
[20] Kyarash Shahriari, and Mana Shahriari, “IEEE Standard Review — Ethically Aligned Design: A Vision for Prioritizing Human Wellbeing with Artificial Intelligence and Autonomous Systems,” 2017 IEEE Canada International Humanitarian Technology Conference (IHTC), Toronto, ON, Canada, pp. 197-201, 2017.
[CrossRef] [Google Scholar] [Publisher Link]
[21] Ethics Guidelines for Trustworthy AI, European Commission, 2019. [Online]. Available: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai
[22] Asilomar AI Principles, Future of Life Institute, 2017. [Online]. Available: https://futureoflife.org/ai-principles/
[23] Human Potential Intelligence, HireVue. [Online]. Available: https://www.hirevue.com/