\newpage \section{Stakeholders} To contextualise and understand the values of the general users of ChatGPT's product, it is important to identify all of our potential stakeholders (we aim to look at least 20 here). From these stakeholders, we distill a named few recurring and shared values that are prioritised across the diverse set. Below is a list of these stakeholders with representative facts about them: \begin{itemize} \item \textbf{Nature of Usage:} Grouping their nature of use to either primary, secondary or tertiary users (based on whether they are users of the interface [direct], they have others use it for them [indirect], or if they are affected collaterally [peripheral]) \item \textbf{Concerns:} Their concerns and potential pain points when using the AI Models \item \textbf{Benefits:} Benefits they get from the product, to understand the value GPT offers to them \item \textbf{Priorities:} Their priorities when using GPT are based on the concerns and benefits \end{itemize} \hspace*{-1cm} \begin{longtable}{|p{0.13\textwidth}|p{0.1\textwidth}|p{0.25\textwidth}|p{0.25\textwidth}|p{0.20\textwidth}|} \hline \textbf{Stakeholder} & \textbf{Nature of usage} & \textbf{Concerns} & \textbf{Benefits} & \textbf{Priorities} \\ \hline \endfirsthead \hline \textbf{Stakeholder} & \textbf{Nature of usage} & \textbf{Concerns} & \textbf{Benefits} & \textbf{Priorities} \\ \hline \endhead Professor & Direct/Indirect & A professor's primary concerns lie in their students' potential misuse of GPT to circumvent actual learning and effort. They also worry about the reliability and accuracy of GPT's claims and summarisations. & Chat GPT would allow a professor to form a study plan and summarise learning material for teaching students. & Reliability, Transparency, Accountability of users, Content Ownership, Trust \\ \hline (Under Graduate/ Post Graduate) Student & Direct & A student would be concerned about the program's reliability in providing accurate answers and the copyright surrounding relevant material. & Students may benefit from GPT's ability to check work and write reports. They could also use GPT chatbots for tutoring. & Reliability, transparency, Content Ownership \\ \hline Researchers & Direct & Researchers would be concerned with the potential for GPT to hallucinate data and draw incorrect conclusions. They would also be concerned with any information GPT provides to their work being accurate and ethically sourced. & A researcher would use GPT to summarise data and statistics. They could also use it to hypothesise and speculate on potential trends. & Reliability, transparency, Fairness, Ownership, Freedom from bias \\ \hline Research Director & Indirect & A research director would be concerned with their researchers' use of GPT and its security. He would also be concerned with the potential biases created by GPT. & A research director may use GPT to review research results and reports to ensure accuracy in findings and conclusions. & Reliability, Transparency, Fairness, Trust \\ \hline Project supervisor & Indirect & A project supervisor would be concerned with the fairness GPT employs when performing admin work and their accountability when using a tool like it. & A Project supervisor would benefit from GPT to summarise information and staff details. & Accountability of user, Fairness, Inclusiveness, Content Ownership \\ \hline PhD student & Direct & A PhD student would be concerned with the accuracy and reliability of GPT's findings and any potential hallucinations it may create. They would also be concerned with the sourcing and copyrighting of any information it provides. & A PhD student may use GPT to summarise large data banks and readings, and get corrections on their Thesis. & Transparency, Reliability, Content Ownership \\ \hline Journalist & Direct & A Journalist would be concerned with the accuracy of GPT when collating an article and any bias it may create in presenting news. & A Journalist would find value in GPT for simplifying complex ideas into a form understandable to the masses. & Fairness, Privacy and security, Human Welfare \\ \hline Educator & Direct & An educator would be concerned with any misinformation GPT may provide when forming a lesson plan. & An educator may also want to use GPT to simplify ideas and draft a teaching plan to teach hard concepts more easily. & Reliability, transparency, Accountability from users \\ \hline Enthusiast group & Direct & An enthusiast group would be concerned with the security protocols protecting member data and the ability for members of all skill levels to get involved with the AI. & An enthusiast group may use GPT to simplify and share information amongst members. & Inclusiveness, Reliability, Universal usability \\ \hline Lawyer & Direct & Lawyers would be primarily concerned with any hallucinations GPT may create as well as the security it has when handling potentially confidential information. & Lawyers would have an interest in using GPT to condense large numbers of case files together to draw out important ones to their case. They could also use AI to summarise arguments and prompt new ideas. & Reliability, Privacy and security, Human Welfare \\ \hline Activist & Indirect & An activist would be concerned about the moral use of GPT and its potential in harming humans. They would also be concerned with the potential for GPT to be used to steal or misuse other people's work. & An activist would benefit from GPT's use in analysing usage statistics and summarising researched data into a more digestible form. & Transparency, Accountability from users, Human Welfare \\ \hline Writer & Indirect & A writer would be concerned about the ownership of work created by GPT and the potential of GPT being used to steal their work without permission. & A writer would benefit from using GPT to prompt ideas, check writing and find faults in their drafts. & Reliability, Universal usability \\ \hline Security personnel & Indirect & Security personnel would be concerned with the potential leaking of information through AI systems. They would also be concerned with the improvements in deciphering bots using AI. & Security personnel would need to know the potential risks of using GPT and how it could be compromised. They may also try to use GPT for complex encryption and ciphering. & Privacy and security, Accountability, Human Welfare \\ \hline Education Board member & Indirect & A board member would be concerned about using GPT to plagiarise work and cause legal trouble. & A board member may use GPT for administrative tasks and busy work. GPT would be useful to summarise documentation and act as an assistant. & Privacy and security, Human Welfare, Freedom from bias \\ \hline Editor & Direct & Editors would be concerned with GPT making incorrect assumptions or altering text incorrectly. They would also be concerned with the potential of unknowingly being given GPT-created work that could lead to copyright or misinformation troubles. & An editor could use GPT to edit and review transcripts. & Content Ownership, Accountability from users \\ \hline IT Engineer & Direct & An IT engineer would be concerned with GPT giving poor code with hard-to-find bugs. They need to understand what a piece of code does, and if too much is generated by GPT, they may not be able to. & An IT engineer would use GPT to write and check the code. Code reviews would help them debug code, and GPT could also help them communicate their work. & Privacy and security, Transparency, Ownership and Property \\ \hline Statistician & Direct & A statistician would be concerned with the potential of GPT hallucinating data and creating biased statistics. & A statistician would use GPT to comb through data and analyse it. GPT can create trends and predictions based on large datasets. & Transparency, Freedom from bias \\ \hline Investors & Indirect & Investors would be concerned by the potential legal ramifications of GPT being used to generate content and any legal infringements users make. & Investors would be interested in the growth of GPT as a product and the stock growth of its parent companies. & Privacy and security, Content Ownership, Trust \\ \hline OpenAI & Indirect & OpenAI would be concerned by the potential legal ramifications of GPT being used to generate content and any legal infringements users make. & OpenAI would be interested in the growth of GPT as a product, and its profits for staff and investors. & Privacy and security, Human Welfare, Content Ownership, Trust \\ \hline Hackers & Indirect & Hackers would be concerned by the potential ciphering and encryption created by GPT. & Hackers are interested in finding flaws and vulnerabilities in data and software created in GPT. They would also use this to engineer human-like personas to trick others into revealing data & Privacy and security, Transparency, Universal Usability, Reliability, and Accuracy \\ \hline \caption{Stakeholder Analysis} \label{tab:stakeholders} \end{longtable}