User:EvaZhuu/Report

Generative AI tool pairs are everywhere in the current society and have started to introduce them in many platforms and online organizations. For the Wikimedia Foundation to use it to help effectively disseminate high-quality content, it must ensure the proper use of generative AI and analyze its pros and cons in order to be prepared to deal with it, and this report will present recommendations to the WMF for managing the impact of generative AI tools on Wikipedia's online community.

I think using generative AI tools to help guide new editors into the Wikipedia community is a great way to do this. As a new Wikipedia editor, I often face challenges with understanding certain editing formats. While Wikipedia provides detailed instructions, it often required searching and jumping through many pages to find them. The Teahouse research article is a great example of the importance of personalized guidance for novice editors (Morgan, Halfaker (2018)), It is a platform for novice editors to engage in Q&A and for experienced editors to interact with and mentor novice editors. Drawing on this research, I think it's possible to turn the Generate AI tool into a role for these experienced editors. I feel that having real-time access to an AI tool to make corrections as edits are made, providing direct links to relevant examples to view, or having an AI tool next to the editor to ask questions would greatly increase the productivity of editors and ensure that some of the best editors are motivated to patiently continue to do quality work for Wikipedia. We can also use AI tools to help reduce the workload of Wikipedia staff, for example by improving the efficiency of reviewing articles. I noticed that many high-quality articles remain at the stub level due to delayed reviews, impacting user experience and editor motivation In this regard, I believe that AI tools can be used to automatically detect whether an article can be upgraded to a higher level every time an editor edits it, which will be more efficient and give editors a sense of accomplishment.

While AI-generating tools offer significant advantages in helping novice editors integrate into the Wikipedia community, there are some potential drawbacks and risks associated with their use. First, over-reliance on AI may lead to a decrease in editors' self-learning skills. In my editing experience, while it sometimes took time to look up editing guidelines between pages, this process of self-directed learning helped me gain a deeper understanding and mastery of Wikipedia's editing rules. If generative AI tools rely too heavily on real-time hints and suggestions, it may result in less opportunity for novice editors to explore and learn on their own, or these newly learned skills may be easily forgotten, affecting the development of their long-term editing skill. So my suggestion would be to let new editors use the AI tool in the early days, giving them detailed and real-time guidance, but then when the editors have gained some experience, the guidance from the AI tool can be changed and they can be encouraged to go for self-directed learning, so they can actually turn some of the skills into their editing abilities.

Secondly, the AI tool may not be so humanized because some processing and guidance are stereotyped, and if it doesn't know how to adapt, it may lead to the fact that it can't suggest the right guidance to the editors according to the different situations when some new issues arise, and then mislead the editors to act in a way that doesn't conform to the rules of Wikipedia. So I think more humanized interactions can be introduced into AI tools so that the AI can communicate with editors according to different issues, the user's editing style and level, and so that the AI can gain adaptability to deal with complex issues through technology.

Third, the AI tool may be biased and wrong because it works based on some existing data and algorithms, and some developers may not have a broad enough view, so some content may be biased. So on top of that I think it's possible for AI tools to do more to input more diverse data and try to be able to embrace the diversity of different cultures and perspectives so that the content that helps the editors stays neutral.

Finally, allowing editors to use AI-generating tools too much may reduce the affective commitment of members. In my current editing process, I think that some editors may go to the “talk” page of a particular Wikipedia page to communicate and help each other when they encounter difficulties, which will help deepen the intimacy between editors and thus promote their motivation to edit for Wikipedia in the long run. However, if the AI tool interferes too much with them, and they can get guidance directly from the AI tool, then it will reduce the communication between editors to some extent, thus reducing their commitment (Kraut & Resnick, 2012). My suggestion on this issue is that a hybrid model could be used, I think AI guidance is necessary but we could encourage and at the same time allow editors to find out that they can communicate this issue in more depth with other editors, for example the AI could help to automate grammatical or misspelled words, but editors at the same time could also send these issues to a specific place of communication to see if there are other forms of expression to help express it better.

In conclusion, generative AI tools have had a mixed impact on Wikipedia's online community, but sensible planning of usage will go a long way towards bringing WMF closer to achieving its goals. We can use different ways of guiding editors at different times to make the AI tools more user-friendly, with more diversity data, and to guide them to get a mix of AI tools and editors interacting with each other to challenge the dilemmas they encounter.