跳到主要內容

簡易檢索 / 詳目顯示

回結果列表
研究生： 舒天宓
Thémis Saquet
論文名稱： 談話型AI如何扭曲與塑形全新的使用者體驗？
How do biases in conversational artificial intelligence distort and shape a new user experience?
指導教授： 莊皓鈞
Chuang, Howard
口試委員: 莊皓鈞
Chuang, Howard
周彥君
Chou, Yen-Chun
學位類別： 碩士
Master
系所名稱： 商學院 - 國際經營管理英語碩士學位學程(IMBA)
International MBA Program College of Commerce(IMBA)
論文出版年： 2023
畢業學年度： 111
語文別： 中文
論文頁數： 61
中文關鍵詞： 偏見 、對話式人工智能 、對話式營銷 、用戶體驗
外文關鍵詞： Bias, Conversational Artificial Intelligence, Conversational Marketing, User Experience
相關次數： 點閱：217  下載：16 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • In our post-COVID-19 societies, more and more consumers rely on conversational AI such as voice assistants or chatbots to perform any kind of task, from asking for the weather to having personal conversations. Companies have seized on this demand to continue developing their conversational AI to create an ever-better user experience. However, the racial or sexist biases implemented in these AIs distort the original user experience, sometimes creating a new one depending on when the bias is implemented. We will try to analyze the effect of these different biases on the user experience and know how they can distort the user experience, especially depending on the moment when these biases appear. To do so, we will analyze the biases in the case of voice assistants and interactive social chatbots through a case study between XiaoIce and MicrosoftTay. We will analyze the appearance of these biases and their effects using the three-stage framework for artificial intelligence in marketing. The main conclusions are that racial biases, mainly embedded because of insufficiently diverse data and engineers, and gender biases, tend to reinforce the structural inequalities that affect our societies. By reinforcing these inequalities, the user experience is negatively impacted in terms of accessibility, representation, and experience conveyed by the use of the product.

    1 INTRODUCTION 1
    2 VOICE ASSISTANTS 8
    2.1 RACIAL BIAS IN VOICE ASSISTANTS 9
    2.1.1 Marketing Research: How the critical step of data collection may influence the user experience from the very conception of the product 9
    2.1.2 Marketing Strategy: How bias becomes a barrier to relevant positioning 14
    2.1.3 An empirical measurement of biases to be corrected 16
    2.1.4 Marketing Action: How a greater personalization through more diverse data and workforce is the way to go for companies 19
    2.2 GENDER BIAS IN VOICE ASSISTANTS 21
    2.2.1 Marketing Research and Strategy: How has a persisting historical bias made its way to these critical steps around the user experience? 21
    2.2.2 Marketing Action: How better marketing practices can help stop these everlasting stereotypes? 27
    3 CHATBOT 29
    3.1 CONVERSATIONAL MARKETING 30
    3.1.1 The benefits of unbiased conversational marketing 32
    3.1.2 What strategies are used for conversational marketing? 39
    3.1.3 Creating a bias-free user experience thanks to conversational marketing by leveraging AI in the different stages of marketing 40
    3.1.3.1 Marketing Research 40
    3.1.3.2 Marketing Strategy 42
    3.1.3.3 Marketing Action 43
    3.2 THE STUDY CASE OF INTERACTIVE SOCIAL CHATBOTS: HOW LETTING USERS IMPLEMENT THOSE BIASES CAN LEAD TO A TERRIBLE USER EXPERIENCE 44
    4 CONCLUSION 52
    5 BIBLIOGRAPHY 56
    6 ADDITIONAL RESOURCES 61

    Bibliography
    1. De Cosmo, L. (2022) Google Engineer claims AI chatbot is sentient: Why that matters, Scientific American. Scientific American. Available at: https://www.scientificamerican.com/article/google-engineer-claims-ai-chatbot-is-sentient-why-that-matters/
    2. Pourquoi l'IA Conversationnelle ? IBM. Available at: https://www.ibm.com/fr-fr/topics/conversational-ai#:~:text=L'intelligence%20artificielle%20(IA),auxquelles%20les%20utilisateurs%20peuvent%20parler
    3. Le fonctionnement des assistants vocaux en 5 étapes (2020) CNIL. Commission nationale de l'informatique et des libertés. Available at: https://www.cnil.fr/fr/le-fonctionnement-des-assistants-vocaux-en-5-etapes#:~:text=Un%20assistant%20vocal%20est%20un,la%20requ%C3%AAte%20d'un%20utilisateur
    4. Qu'est-Ce Qu'un Biais Cognitif ? Définition Biais cognitif (2023) USABILIS. Available at: https://www.usabilis.com/definition-biais-cognitifs/
    5. Spradlin, L.K. (2012) Diversity matters: Understanding diversity in schools. Belmont, CA: Wadsworth, Cengage Learning.
    6. Braswell, P. (2022) This is the difference between racism and racial bias, Fast Company. Available at: https://www.fastcompany.com/90796690/this-is-the-difference-between-racism-and-racial-bias
    7. Eliminating gender bias in conversational AI: Strategies for fair and Inclusive Conversations (2023) Aivo. Available at: https://www.aivo.co/blog/eliminating-gender-bias-in-conversational-ai-strategies-for-fair-and-inclusive-conversations
    8. User experience (2020) What is User Experience? | Definition and Overview. Available at: https://www.productplan.com/glossary/user-experience/
    9. What is user experience (UX) design? (2022) The Interaction Design Foundation. Interaction Design Foundation. Available at: https://www.interaction-design.org/literature/topics/ux-design
    10. Huang, M.-H. and Rust, R.T. (2020) A strategic framework for artificial intelligence in marketing - journal of the Academy of Marketing Science, SpringerLink. Springer US. Available at: https://link.springer.com/article/10.1007/s11747-020-00749-9
    11. Edison Research (2020) The smart audio report 2020 from NPR and Edison Research, Edison Research. Edison Research http://www.edisonresearch.com/wp-content/uploads/2014/06/edison-logo-300x137.jpg. Available at: https://www.edisonresearch.com/the-smart-audio-report-2020-from-npr-and-edison-research/
    12. Transcript: Ai from above (EP3) (2022) The Internet Health Report 2022. Available at: https://2022.internethealthreport.org/transcript-ai-from-above-ep3/
    13. Harwell, D. (2018) The accent gap: How Amazon's and Google's smart speakers leave certain voices behind, The Washington Post. WP Company. Available at: https://www.washingtonpost.com/graphics/2018/business/alexa-does-not-understand-your-accent/
    14. Sidnell, J. (2019). African American Vernacular English. [online] Hawaii.edu. Available at: https://www.hawaii.edu/satocenter/langnet/definitions/aave.html.
    15. Koenecke, A. et al. (2020) Racial disparities in automated speech recognition | PNAS, Proceedings of the National Academy of Sciences, PNAS. Available at: https://www.pnas.org/doi/10.1073/pnas.1915768117
    16. Online resources for African American language (no date) CORAAL | Online Resources for African American Language. Available at: https://oraal.uoregon.edu/coraal
    17. Lloreda, C.L. (2020). Speech Recognition Tech Is Yet Another Example of Bias.
    Scientific American. Available at: https://www.scientificamerican.com/article/speech-recognition-tech-is-yet-another-example-of-bias/
    18. Cantone, J.A., Martinez, L.N., Willis-Esqueda, C. and Miller, T. (2019). Sounding guilty: How accent bias affects juror judgments of culpability. Journal of Ethnicity in Criminal Justice, 17(3), pp.228–253. doi:https://doi.org/10.1080/15377938.2019.1623963
    19. Baquiran, C.L.C. and Nicoladis, E. (2019). A Doctor’s Foreign Accent Affects Perceptions of Competence. Health Communication, 35(6), pp.726–730. doi:https://doi.org/10.1080/10410236.2019.1584779
    20. Baraniuk, C. (2022). Why your voice assistant might be sexist. [online] www.bbc.com. Available at: https://www.bbc.com/future/article/20220614-why-your-voice-assistant-might-be-sexist
    21. CNN, B.B.G. (n.d.). Why computer voices are mostly female. [online] CNN. Available at: https://edition.cnn.com/2011/10/21/tech/innovation/female-computer-voices/
    22. Gizmodo. (n.d.). No, Women’s Voices Are Not Easier to Understand Than Men’s Voices. [online] Available at: https://gizmodo.com/no-siri-is-not-female-because-womens-voices-are-easier-1683901643
    23. Halo· (2023). How many copies did Halo sell? — 2023 statistics | LEVVVEL. [online] levvvel.com. Available at: https://levvvel.com/halo-statistics/#:~:text=The%20Halo%20series%20has%20sold%2081%20million%20copies.&text=The%20series%20went%20from%2065
    24. Franceinfo. (2020). Nouveau monde. Pourquoi les prénoms de Siri et d’Alexa pour des assistants vocaux ? [online]
    Available at: https://www.francetvinfo.fr/replay-radio/nouveau-monde/nouveau-monde-pourquoi-les-prenoms-de-siri-et-d-alexa-pour-des-assistants-vocaux_4045705.html
    25. Gizmodo. (2015). Why Is My Digital Assistant So Creepy? [online] Available at: https://gizmodo.com/why-is-my-digital-assistant-so-creepy-1682216423
    26. World Economic Forum. (n.d.). Hey Siri, you’re sexist, finds UN report on gendered technology. [online] Available at: https://www.weforum.org/agenda/2019/05/hey-siri-youre-sexist-finds-u-n-report-on-gendered-technology/
    27. www.ibm.com. (n.d.). Qu’est-ce qu’un agent conversationnel ? | IBM. [online] Available at: https://www.ibm.com/fr-fr/topics/chatbots
    28., 29. & 30. Salesforce.com. (n.d.). State of the Connected Customer Report. [online] Available at: https://www.salesforce.com/resources/research-reports/state-of-the-connected-customer/.
    31. Drift. (2021). 2021 State of Conversational Marketing. [online] Available at: https://www.drift.com/books-reports/conversational-marketing-trends/#state-of-convo-marketing.
    32. Weizenbaum, J. (1966). ELIZA---a computer program for the study of natural language communication between man and machine. Communications of the ACM, [online] 9(1), pp.36–45. doi:https://doi.org/10.1145/365153.365168.
    33. Colby, K.M., Weber, S. and Hilf, F.D. (1971). Artificial Paranoia. Artificial Intelligence, 2(1), pp.1–25. doi:https://doi.org/10.1016/0004-3702(71)90002-6.
    34. Wallace, R.S. (2007). The Anatomy of A.L.I.C.E. Parsing the Turing Test, [online] pp.181–210. doi:https://doi.org/10.1007/978-1-4020-6710-5_13.
    35. Zhou, L., Gao, J., Li, D. and Shum, H.-Y. (2020). The Design and Implementation of XiaoIce, an Empathetic Social Chatbot. Computational Linguistics, 46(1), pp.53–93. doi:https://doi.org/10.1162/coli_a_00368.
    36. Quach, K. (n.d.). Microsoft chatbots: Sweet XiaoIce vs foul-mouthed Tay. [online] www.theregister.com. Available at: https://www.theregister.com/2016/09/29/microsofts_chatbots_show_cultural_differences_between_the_east_and_west/
    37. Microsoft chatbot is taught to swear on Twitter. (2016). BBC News. [online] 24 Mar. Available at: https://www.bbc.com/news/technology-35890188.
    38. Hunt, E. (2018). Tay, Microsoft’s AI chatbot, gets a crash course in racism from Twitter. [online] The Guardian.Available at: https://www.theguardian.com/technology/2016/mar/24/tay-microsofts-ai-chatbot-gets-a-crash-course-in-racism-from-twitter.
    39. Explains, K. (2020). Remembering Microsoft’s Chatbot disaster. [online] Medium. Available at: https://uxplanet.org/remembering-microsofts-chatbot-disaster-3a49d4a6331f
    40. Editor (2016). XiaoIce Vs. Tay: Two A.I. Chatbots, Two Different Outcomes - Sampi. co. [online] Available at: https://sampi.co/chinese-chatbot-xiaoice-vs-tay/
    41. & 42. Miller, K.W., Wolf, M.J. and Grodzinsky, F.S. (2017). Why We Should Have Seen That Coming. ORBIT Journal, 1(2). doi:https://doi.org/10.29297/orbit.v1i2.49
    43. Leetaru, K. (n.d.). How Twitter Corrupted Microsoft’s Tay: A Crash Course In the Dangers Of AI In The Real World. [online] Forbes. Available at: https://www.forbes.com/sites/kalevleetaru/2016/03/24/how-twitter-corrupted-microsofts-tay-a-crash-course-in-the-dangers-of-ai-in-the-real-world/?sh=ffeebf926d28
    44. https://plus.google.com/+UNESCO (2021). Problème de fille : briser les préjugés dans l’IA. [online] UNESCO. Available at: https://fr.unesco.org/girltrouble
    45. Zendesk. (n.d.). 7 ways to reduce bias in conversational AI. [online] Available at: https://www.zendesk.tw/blog/7-ways-reduce-bias-conversational-ai/#georedirect
    46. SoundHound. (n.d.). How to Overcome Cultural Bias in Voice AI Design. [online] Available at: https://www.soundhound.com/resources/how-to-overcome-cultural-bias-in-voice-ai-design/
    47. Dotan, R. (2023) Example of gender bias in CHATGPT, Ravit Dotan. Available at: https://www.ravitdotan.com/post/example-of-gender-bias-in-chatgpt
    48. Shafik and Case (eds.) (2022) Advances in manufacturing technology XXXV, IOS Press. Available at: https://www.iospress.com/catalog/books/advances-in-manufacturing-technology-xxxv
    49. Kotek, H. (2023) Doctors can’t get pregnant and other gender biases in chatgpt, Doctors can’t get pregnant and other gender biases in ChatGPT. Available at: https://hkotek.com/blog/gender-bias-in-chatgpt/
    Additional resources
    1. Wheeler, D.R. (2014). Why are Cortana and Siri female? [online] CNN. Available at: https://edition.cnn.com/2014/04/04/opinion/wheeler-cortana-siri/
    2. Abby Ohlheiser, covering digital culture (n.d.). Trolls turned Tay, Microsoft’s fun millennial AI bot, into a genocidal maniac. [online] Washington Post. Available at: https://www.washingtonpost.com/news/the-intersect/wp/2016/03/24/the-internet-turned-tay-microsofts-fun-millennial-ai-bot-into-a-genocidal-maniac/
    3. Mathur, V., Stavrakas, Y. and Singh, S. (2016). Intelligence analysis of Tay Twitter bot. [online] ResearchGate. Available at: https://www.researchgate.net/publication/316727714_Intelligence_analysis_of_Tay_Twitter_bot
    4. Nouri, S. (n.d.). Council Post: The Role Of Bias In Artificial Intelligence. [online] Forbes. Available at: https://www.forbes.com/sites/forbestechcouncil/2021/02/04/the-role-of-bias-in-artificial-intelligence/?sh=3d43f9f7579d
    5. UN News. (2019). Are robots sexist? UN report shows gender bias in talking digital tech. [online] Available at: https://news.un.org/en/story/2019/05/1038691
    6. West, M., Kraut, R. and Chew, H.E. (2019). I’d blush if I could: closing gender divides in digital skills through education. Available at: https://unesdoc.unesco.org/ark:/48223/pf0000367416.page=1

    QR CODE
    :::