Thinking about my future job as a teacher, I believe AI becomes too much, for example, when AI-composed homework and papers are corrected by AI. By that, I don’t mean that AI can’t be used as an assistive tool at all, but it is about the extent to which one uses it. Personally, I think it becomes too much when you have it compose or correct whole assignments without reflecting on the topics and outputs yourself. Having AI tools compose whole papers is clearly too much because, as we have discussed in class, and as is argued by Tang et al. (2023), as of now, AI tools are non-legal entities, meaning they cannot be granted authorship and, thus, cannot take responsibility for their outputs. Additionally, most LLMs still do a poor job of referencing sources, therefore running risk of plagiarizing (Tam et al., 2023; Tang et al., 2023).

Overall, I feel clear guidelines, for example, on how to cite which kind of AI use, would be necessary to increase transparency. Nevertheless, as is the case with other tools, like Grammarly or even ghostwriters, guidelines would still not guarantee responsible and appropriate use. After all, it will still be up to the author to declare AI use honestly, while the reader has to form their own judgment of a text.

Here’s an example from Austria, which highlights the lack of knowledge of, as well as guidelines for, AI use. In 2015, the “Vorwissenschaftliche Arbeit”, a “pre-scientific paper” all students graduating from high school were required to write during their last school year, was introduced. Because of the fast development of AI and teachers’ inability to assess whether students’ work was written by AI or the students themselves, the Ministry of Education decided to remove this requirement for graduation. Many politicians have criticized this decision, stating that completely abandoning the pre-scientific paper should not be the solution. Rather, it should be discussed how AI can be incorporated appropriately to enhance student learning.

Sources:

Die Presse (Austrian Newspaper). Vorwissenschaftliche Arbeit soll abgeschafft werden.https://www.diepresse.com/18532278/vorwissenschaftliche-arbeit-soll-abgeschafft-werden

Tam, W., Huynh, T., Tang, A., Luong, S., Khatri, Y., & Zhou, W. (2023). Nursing education in the age of artificial intelligence powered Chatbots (AI-Chatbots): Are we ready yet? Nurse Education Today, 129, 105917. https:// doi. org/ 10. 1016/j. nedt. 2023. 105917

Tang, A., Li, K-K., Kwok, K. O., Cao, L., Luong, S. & Tam, W. (2024). The importance of transparency: Declaring the use of generative artificial intelligence (AI) in academic writing. Journal of Nursing Scholarship, 56, 314–318. https://doi.org/10.1111/jnu.12938 15475069, 2024, 2, Downloaded from https://sigmapub

Experimenting with creative writing in LLMs was an insightful task. However, I would like to mention that I basically never do anything related to creative writing (maybe kind of a sad insight?), so I feel I might not be a good judge of an LLM’s creative output.

For the creative writing task, I tried several prompts and especially liked the output I got from the following prompt in ChatGPT and Perplexity (see attached Word document for the complete output):

• Generate a poem in the style of Edgar Allan Poe about the feeling of losing Wi-Fi when out camping in a forest.

Overall, I probably would not notice that the poems were actually written by an LLM. Maybe I could tell, if I were more familiar with Edgar Allan Poe’s writing style, but I can only remember his poem The Raven, which I read some years ago (see here, if you’re interested:https://www.btboces.org/Downloads/7_The%20Raven%20by%20Edgar%20Allen%20Poe.pdf) – so this was my main point of comparison.

Both LLMs did a really good job at capturing the dark, melancholic writing style of the author. Also, both poems used first-person narration, which is common for Poe.

Comparing the two LLMs, I think I liked Perplexity’s output more than ChatGPT’s, just because it flows better in my opinion. However, the stanzas are maybe a little short for a typical Poe poem, and I found one word of which I’m not sure whether it is made up or actually exists (maybe it is an archaic word) – embered.

Regarding the ChatGPT output, I noticed that some words were capitalized mid-sentence, and I didn’t know why that was the case. When I asked ChatGPT, it stated that this is typical of Poe’s writing style, so I compared it to his work The Raven. Indeed, I was able to find some words that were capitalized mid-sentence; however, this stylistic device was used very sparingly only, so I think that is a good example of how LLMs overuse a certain feature after detecting a pattern.

I think the outputs from both LLMs were very creative. In line with Boden’s (Arriagada & Arriagada-Bruneau, 2022) definition, I think being creative necessitates that an (art)work is novel, surprising, and valuable. Both poems succeeded in that they identified the author’s writing style (more or less) correctly and then applied it to a new context. As we also discussed in class, value, of course, is very subjective, as different people, cultures, etc., might have different understandings of what is valuable.

I was surprised by the studies showing that people appreciate artwork less when they know it was produced by AI (Arriagada & Arriagada-Bruneau, 2022). Similar to the example of the invention of photography, as we also briefly discussed in class, I think AI creativity will not replace but change human-made art. Nevertheless, I see that there is a bigger issue with AI than, for instance, the introduction of photography, given that AI is trained on people’s original ideas (predominantly without asking them for consent and permission), so I feel this is an issue that needs urgent attention.

Overall, I think the experiments were fun and a good way to get a feeling for LLMs’ potential to be used in creative work. I think my idea of creativity has always resembled the definition by Boden (Arriagada & Arriagada-Bruneau, 2022), but I couldn’t pin it down, so now I can better put into words what I consider creative.

Sources:

Arriagada, L., & Arriagada-Bruneau, G. (2022). AI’s Role in Creative Processes: A Functionalist Approach. Odradek. Studies in Philosophy of Literature, Aesthetics, and New Media Theories, 8 (1), 77-110.

Poe, Edgar, A. 1845. The Raven. Online https://www.btboces.org/Downloads/7_The%20Raven%20by%20Edgar%20Allen%20Poe.pdf

This week’s reading, videos, and discussions have improved my understanding of prompting in the context of large language models. I learned that several “prompt patterns” exist that can be applied to improve an LLM’s output quality (see, section III. of White et al., 2023).

In the past, I didn’t really think about how to formulate my prompts. I just went by my “gut feeling,” and that worked pretty well. This is probably partly because, as mentioned by Mollick (2023), the term prompt engineering, as it is often defined, is overly complex and probably more appropriate for use in the context of computer programming. However, with LLMs always advancing, their use has become more interactive, allowing for more human-like conversations, for example, by means of iterative prompting (see, e.g., the Mollicks, 2023, video on prompting we watched for our class on April 2), rather than relying on very specific input to provide useful output. Nevertheless, I think knowing about certain prompt patterns is useful for reaching a qualitative output more quickly.

For the prompting experiment, I tried out the flipped interaction pattern combined with the persona pattern. I used the following prompt to compare the outputs from ChatGPT vs. Perplexity:

“Help me plan my California trip. Imagine you are a travel agent. Ask me as many questions as you need to provide me with a 1-week holiday tailored to my interests.”

I think asking an LLM to take on the persona of a travel agent can be useful for anyone planning a trip and feeling a little overwhelmed by the task – especially when they don’t know much yet about the destination they want to travel to. I am quite familiar with ChatGPT but had never used Perplexity before. I was surprised how good the questions and responses were in both of the LLMs. The questions I was asked overlapped a lot, and both LLMs suggested similar itineraries. ChatGPT asked me 20 questions, while Perplexity asked 21. Overall, I think ChatGPT has a more user-friendly design – it does a really good job of using emojis or bold-facing important parts, for example. Perplexity seemed a little more “professional” and straightforward in its tone. I also liked that Perplexity included a section called “Other considerations” to allow me to add anything it might have forgotten to ask.

Sources:

• Mollick, E. (2023). A guide to prompting AI for what you want. One Useful Thing. https://www.oneusefulthing.org/p/a-guide-to-prompting-ai-for-what
• Wharton School. (2023). Practical AI for instructors and students part 3: Prompting AI. YouTube. https://www.youtube.com/watch?v=wbGKfAPlZVA
• White, J., Fu, Q., Hays, S., Sandborn, M., Olea, C., Gilbert, H., Elnashar, A., Spencer-Smith, J., & Schmidt, D. C. (2023). A prompt pattern catalog to enhance prompt engineering with ChatGPT. arXiv. https://doi.org/10.48550/arXiv.2302.11382

When thinking about the ethical issues involved in AI before this class, the first things that came to my mind used to be the invasion of privacy (e.g., usage of personal data or surveillance systems), displacement of humans in various jobs (e.g., in data entry/analysis), and issues of plagiarism (i.e., using AI tools to create work without giving credit).

This week’s lectures, the class readings, and discussions with peers have all taught me that the ethical issues of AI are far more wide-reaching and complex. What I did not know before, and thus what really struck me, was the fact that one single search on an LLM like ChatGPT takes up an entire bottle of water. (Coincidentally, a friend of mine posted this on her Instagram this week: https://www.instagram.com/zerowastestore/reel/DHB4s3UyIX0/). As we learned in the article by Bender et al. (2021), the people who suffer the most from the negative environmental consequences of LLMs are usually those who benefit the least from their progress, thus highlighting the intersectionality of ethical issues in LLMs.

After class, I continued my reflection on potential areas of conflict. One area we haven’t really discussed yet but that certainly is of relevance too is the use of AI for providing psycho-social support, such as in psychological counseling and perhaps even as a replacement for social and romantic relationships. This is an area of AI use that surfaced a long time ago in many utopian or dystopian movies (e.g., Her, 2013) but, as a quick online search shows, no longer just exists in movies.

Several articles (e.g., Ping, 2024) have investigated the use of AI technology, such as AI chatbots, in psychological counseling. Scholars have also examined the potential of AI for social and romantic relationships, focusing, for example, on how people can have meaningful relationships with chatbots (e.g., Pan & Mou, 2024). Considering that we live in a world in which individuals seem to increasingly prioritize self-fulfillment over (romantic) relationships, this area of AI seems particularly interesting to me.

With the last ChatGPT update, I have noticed that the chatbot has become more personal, for instance, by using smileys and asking questions like, “Let me know if I can help you with x or y….”

• What do you think about the use of AI in areas like counseling or as a replacement for or addition to social relationships?
• Do you use neutral or friendly language when engaging with an AI chatbot?

I’m looking forward to hearing your opinions on this topic! 🙂

Sources:

• Bender, E. M., McMillan-Major, A., Gebru, T., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’21), 610–623. https://doi.org/10.1145/3442188.3445922
• Her. IMBD. https://www.imdb.com/title/tt1798709/
• Pan, S., & Mou, Y. (2024). Constructing the meaning of human–AI romantic relationships from the perspectives of users dating the social chatbot Replika. Personal Relationships, 31(3), 1090–1112.https://doi.org/10.1111/pere.12572
• Ping, Y. (2024). Experience in psychological counseling supported by artificial intelligence technology. Technology and Health Care, 32(6), 3871–3888. https://doi.org/10.3233/THC-230809

Hello,

I’m Astrid, and I go by the pronouns she/her/hers. I’m enrolled in the teacher education program for the subjects of English and psychology/philosophy/ethics at Vienna University. Here in Wooster, I’m a teaching assistant for German and also attend classes as a student. In my free time, I enjoy baking and going on hikes, and I also love exploring coffeehouses with my family and friends.

Regarding AI, I would like to mention that, as I feel it is becoming ever more important and intertwined with our daily lives, I’d rather learn to use it efficiently and ethically than neglect it altogether.