Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Ethics in AI Security Implications

Ethics in AI Security Implications ETHICS IN AI SECURITY IMPLICATIONS Ian Fish FBCS, Chair of the Information Security Specialist Group, gives some thought to current issues around ethics and AI in the cyber security arena. In the last issue I wrote about security and causation. The fourth issue was that AI unconscious bias? Even the writing of the the ethics of AI. The editor informed me could be used for good or evil and the fi fth algorithms could be subject to unconscious that the theme of this issue is ethics and I was liability and legal issues. The fi nal bias with the developers’ innate beliefs felt as if I’d already shot my bolt. However, issue was the potential e ect about the nature of the problem they Chris Rees (the BCS President) came to my on employment. are trying to solve. To take it to the next rescue in his fi rst day after election when So, how do these issues play out for AI stage, what if there is continuous machine he addressed the Member Groups Spring in the cyber security space? I made some learning while the AI is in operation? Convention on his theme for the year – preliminary observations about the last Even if the need for transparency on ethics – and his six potential issues (about three in the previous issue, so I should like how decisions are reached is achieved AI again). This got me thinking again about to concentrate on the fi rst three issues, initially how can that transparency be AI in security. which are essentially about the design and maintained as the AI learns? I have only The fi rst issue raised was the e ect of implementation of AI systems. touched the surface here, but I recommend intrinsic bias, either in the data set or the If we look at an AI implementation that as AI is implemented in our discipline AI algorithms. The second was the need doing repetitive tasks on large sets of data, you pause to think about the six issues that for transparency in how decisions are we see that it would have to be trained Chris mentioned. made by AI systems, and the third was a fi rst on exemplar large data sets. Are we reminder that correlation does not equal capable of creating such data sets without Security highlights: p42-43: Ian Edwards asks what p44-45: John Mitchell asks p36-37: Christopher Middup GDPR really means for IT and whether ethics or morality suggests key IoT security security teams. figure in white hat hacking. improvements. 34 ITNOW June 2018 Downloaded from https://academic.oup.com/itnow/article-abstract/60/2/34/4999839 by Ed 'DeepDyve' Gillespie user on 20 June 2018 doi:10.1093/itnow/bwy046 ©2018 The British Computer Society Image: getty/ franckreporter http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png ITNow Oxford University Press

Ethics in AI Security Implications

Fish, Ian
ITNow , Volume Advance Article (2) – May 19, 2018
Download PDF
1 page

Loading next page...
 
/lp/ou_press/ethics-in-ai-security-implications-0bTXdBbtMd
Publisher
Oxford University Press
Copyright
© 2018 The British Computer Society
ISSN
1746-5702
eISSN
1746-5710
DOI
10.1093/itnow/bwy046
Publisher site
See Article on Publisher Site

Abstract

ETHICS IN AI SECURITY IMPLICATIONS Ian Fish FBCS, Chair of the Information Security Specialist Group, gives some thought to current issues around ethics and AI in the cyber security arena. In the last issue I wrote about security and causation. The fourth issue was that AI unconscious bias? Even the writing of the the ethics of AI. The editor informed me could be used for good or evil and the fi fth algorithms could be subject to unconscious that the theme of this issue is ethics and I was liability and legal issues. The fi nal bias with the developers’ innate beliefs felt as if I’d already shot my bolt. However, issue was the potential e ect about the nature of the problem they Chris Rees (the BCS President) came to my on employment. are trying to solve. To take it to the next rescue in his fi rst day after election when So, how do these issues play out for AI stage, what if there is continuous machine he addressed the Member Groups Spring in the cyber security space? I made some learning while the AI is in operation? Convention on his theme for the year – preliminary observations about the last Even if the need for transparency on ethics – and his six potential issues (about three in the previous issue, so I should like how decisions are reached is achieved AI again). This got me thinking again about to concentrate on the fi rst three issues, initially how can that transparency be AI in security. which are essentially about the design and maintained as the AI learns? I have only The fi rst issue raised was the e ect of implementation of AI systems. touched the surface here, but I recommend intrinsic bias, either in the data set or the If we look at an AI implementation that as AI is implemented in our discipline AI algorithms. The second was the need doing repetitive tasks on large sets of data, you pause to think about the six issues that for transparency in how decisions are we see that it would have to be trained Chris mentioned. made by AI systems, and the third was a fi rst on exemplar large data sets. Are we reminder that correlation does not equal capable of creating such data sets without Security highlights: p42-43: Ian Edwards asks what p44-45: John Mitchell asks p36-37: Christopher Middup GDPR really means for IT and whether ethics or morality suggests key IoT security security teams. figure in white hat hacking. improvements. 34 ITNOW June 2018 Downloaded from https://academic.oup.com/itnow/article-abstract/60/2/34/4999839 by Ed 'DeepDyve' Gillespie user on 20 June 2018 doi:10.1093/itnow/bwy046 ©2018 The British Computer Society Image: getty/ franckreporter

Journal

ITNow – Oxford University Press

Published: May 19, 2018

There are no references for this article.