admn-233-assignment-2

pdf

School

New York University *

*We aren’t endorsed by this school

Course

MISC

Subject

Computer Science

Date

Jan 9, 2024

Type

pdf

Pages

5

Uploaded by jupyter83

Report
Studocu is not sponsored or endorsed by any college or university ADMN 233 Assignment 2 Writing in Organizations (Athabasca University) Studocu is not sponsored or endorsed by any college or university ADMN 233 Assignment 2 Writing in Organizations (Athabasca University) Downloaded by Umaima Usman (umaimausman2k@gmail.com) lOMoARcPSD|3866758
[Date] Monarchy Technology CORPORATE COMMUUNICATIONS SPECIALIST Analytical Report Regency III Failure Investigation & Analysis Downloaded by Umaima Usman (umaimausman2k@gmail.com) lOMoARcPSD|3866758
Table of Contents Executive Summary ........................................................................................................................................................ 2 Introduction ..................................................................................................................................................................... 2 Potential Causes and Analysis ........................................................................................................................................ 2 Downloaded by Umaima Usman (umaimausman2k@gmail.com) lOMoARcPSD|3866758
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
  • Access to all documents
  • Unlimited textbook solutions
  • 24/7 expert homework help
Executive Summary Regency III has been responsible for a drop in diversity hires for Eidos, Monarchy’s largest customer. During our investigation we have found two potential causes for this. The data Eidos has given to the algorithm may be biased and therefore Regency produces biased results. Eidos used all their information for the past three years including interviewer notes. These are likely what made the training data biased. The interviewer likely made negative notes on candidates based on their own subconscious bias. A second less probable cause is a possible lack of diversity on the Monarchy design team resulting in potentially biased algorithm. Accuracy scores done during tests were above 96%. This means the issue is caused by something on Eidos’ end and not Monarchy’s. Introduction Eidos Logistics has reported significant issues with Regency III which Monarchy is currently investigating. With the previous version or Regency, Eidos was able to achieve 56% alignment with the diversity criteria. This will be an analytical report regarding the issues of Regency III and Eidos’ claims. It is Monarchy’s goal to figure out what is causing the bias and remedy the issue as fast as possible so companies such as Eidos can continue using Regency and receive the best results possible. We will analyze the results of our investigation and address potential causes and the effects of Regency III. To conclude we will discuss possible actions to immediately remedy the situation and preventative measures Monarchy can take in order to avoid future issues. Potential Causes and Analysis Regency III is responsible for a significant drop in Eidos’ diversity. With Regency II Eidos was able to achieve 56% alignment with their diversity criteria. Following the implementation of Regency III, their alignment dropped to 24%. Eidos has concluded that Regency III is affected by algorithm bias and have stopped using the software to mitigate the damage. After an initial investigation, we have narrowed down the potential causes to a few possible causes of Eidos’ possible experience. 1. Bias within the training data provided to the software. 2. There are issues with the foundation of the software, reflecting the bias of the team who created it. We know Eidos was responsible for providing the training data. They used all their hiring information from the past three years, this included accepted and rejected candidates, resumes, interview transcripts and the interviewers’ personal notes. It is possible this set of data may be biased causing our software to behave in a biased way. The use of the interviewers notes on candidates is likely the cause of the bias as it leaves to much room for personal thoughts and feelings to be taken into account by the algorithm. Downloaded by Umaima Usman (umaimausman2k@gmail.com) lOMoARcPSD|3866758
It is also possible there are issues with the foundation of the software. An algorithm can reflect the bias oof the team that created it. If there are issues with the diversity of Monarchy’s design team, this could result in the creation of a biased algorithm. This is less likely as the previous potential cause due to the 96% accuracy score Regency III received during trials. Conclusion While our investigation into this matter is still ongoing, we can draw a preliminary conclusion and make several preliminary suggestions for next steps. It is most likely the cause the failure of Regency III is the training data. I recommend we remove the current training data, and we work with Eidos to create a new set of data. Ideally this new set of data will not include the personal thoughts of the interviewer. This should increase Eidos’ workplace diversity and allow them to continue use of Regency III. To prevent this issue from arising in the future it would also be wise to collaborate with future clients on creating their training data and provide educational resources on training data to all clients. Downloaded by Umaima Usman (umaimausman2k@gmail.com) lOMoARcPSD|3866758