ChatGPT and college-level science writing: A blessing or a curse?

Serge Onyper

St. Lawrence University, Canton, NY

January 19, 2023

As an instructor who regularly teaches college-level writing, I had a lot of questions when I heard of the emergence of a powerful artificial intelligence ChatGPT two months ago (https://chat.openai.com/auth/login). ChatGPT, a chatbot launched by the company OpenAI, uses natural language to answer a variety of user requests. Among these requests are writing an essay or research paper, summarizing a source, or revising one’s writing, all tasks that I regularly ask my students to do in the psychology and neuroscience courses that I teach at St. Lawrence University.

Naturally, I wanted to know what sort of responses ChatGPT could generate to writing prompts I often use in my courses, and how I should prepare for its use by the students who will undoubtedly catch on very quickly that such a tool can potentially change how they approach writing assignments. Unlike a search engine, ChatGPT relies on a fixed corpus of knowledge (with websites, articles, books, and other sources published up to 2021) to construct its responses.

To put the chatbot to a test, I focused on 2 broad goals. The first goal was to generate material for a student research paper that was a literature review. The second goal was to explore its success with smaller assignments that focus on foundational skills required to produce such a paper.

If you are interested in the conclusions from this exercise, feel free to jump to the end to read them. For a detailed analysis, read on.

Goal 1: Research paper

To address the first goal, I began by picking a topic familiar to me. In the summer of 2021, I wrote and submitted for publication an empirical paper that examined the relationship between videoconferencing technology use and sleep in a sample of college students. While the paper is not yet published (I’m currently incorporating a second study from a broader sample into it), I thought I could use the topic as a prompt for a student research paper. To help me generate materials for this paper, I logged in to the chatbot and asked it several questions.

To preview what I learned from this exercise, it is possible to get the chatbot to produce a decent-quality student paper with sufficient breadth, albeit depth is an issue because the summaries of research studies tend to be 1-2 sentences long. In some cases, that might be sufficient, but in other cases further inquiries to the chatbot are needed to develop the paper further. The biggest drawback to using ChatGPT is its inability to accurately cite its sources: Asking for citations or references produces in-text citations and a references page, but in most cases the references are bogus. Trying to locate the sources ChatGPT claims to be using is often impossible, although the citations themselves look legitimate and, in many cases, the authors listed in those citations have published work in the relevant field.

The questions, with ChatGPT’s responses, are below.

Request: Write a long paper with citations on how exposure to videoconferencing may affect sleep.

The widespread adoption of videoconferencing technology as a result of the COVID-19 pandemic has brought about a significant increase in the amount of time that people spend in front of screens. This has raised concerns about the potential negative effects of videoconferencing on sleep.

Several studies have investigated the effects of screen time in general on sleep. A meta-analysis by Van der Lely et al. (2019) found that excessive screen time is associated with poor sleep quality and shorter sleep duration in both children and adults. The study also found that the negative effects of screen time on sleep are particularly pronounced in children and adolescents.

More recent research has specifically looked at the effects of videoconferencing on sleep. A study by Kushida et al. (2021) found that people who participate in videoconferences for more than 5 hours per day have a significantly higher risk of insomnia and poor sleep quality compared to those who participate in fewer videoconferences. The study suggests that the prolonged exposure to blue light and the cognitive demands of participating in videoconferences may contribute to insomnia and poor sleep quality.

Another study by Bu et al. (2020) found that exposure to blue light in the evening, as well as the cognitive load and social demands of videoconferencing, can lead to a delay in the onset of melatonin production, making it more difficult to fall asleep. The study suggests that the prolonged exposure to blue light and the cognitive demands of participating in videoconferences may lead to insomnia and poor sleep quality.

Moreover, a survey study by Léger et al. (2020) found that people who frequently use videoconferencing experience a higher level of stress, which can negatively affect sleep. The study found that people who use videoconferencing frequently have a higher level of stress, which can negatively affect sleep.

It's important to note that the effects of videoconferencing on sleep may vary depending on individual differences, such as age, sex, and pre-existing medical conditions. Moreover, the time of day that the videoconferencing takes place, the type of videoconferencing (synchronous or asynchronous), and the environment in which the videoconferencing takes place (light, noise, and comfort) are all factors that may moderate the relationship between videoconferencing and sleep.

In conclusion, several studies have found that videoconferencing can negatively affect sleep, particularly with prolonged exposure and high frequency. The prolonged exposure to blue light and cognitive demands of videoconferencing can lead to a delay in the onset of melatonin production and contribute to insomnia and poor sleep quality. The increased stress level caused by the social demands of videoconferencing can also negatively impact sleep. To mitigate these negative effects, it is recommended to limit the amount of time spent on videoconferencing, particularly in the evening, and to use blue-light blocking glasses, dim red lights in the bedroom and to have a comfortable environment for videoconferencing, with adequate light and noise level.

References:

As one can see, the response follows a certain structure. It begins with a brief introduction, then presents evidence in support of several claims (usually 3), then includes a concession, and finishes with a conclusion. The response also incorporates a variety of research sources (but you have to ask to include citations or references in the original question, otherwise source citations are omitted).

While the paper ChatGPT produced is quite brief, this structure constitutes an accepted approach to constructing an evidence-based argument in scientific writing: It is a structure that I encourage students in my classes to use when they conceptualize argumentative writing. Importantly, you can ask the chatbot follow-up questions to give the response depth.

Thus, I took the 3 points/claims produced by the chatbot in the paper above and asked it to generate a separate paper on each. Each of the claims refers to a mechanism that explains why videoconferencing might affect one’s sleep: 1) by increasing screen time; 2) by increasing exposure to blue light; 3) by increasing stress.

My requests and the chatbot’s responses are shown below. What is striking is that, if I were to incorporate these 3 mini-papers into the framework provided by the paper above, I would easily end up with a 5-6 page paper that seemingly utilized 20+ sources to construct a compelling (certainly for student work!) argument that explains why extensive videoconferencing might be detrimental to the quality of one’s sleep. Most of the sources cited are not of actual papers, but would a student know it? And, further, would we know it without locating and examining each cited source? The chatbot's argument seems to hold water, and it took me no time at all to come up with questions and generate responses. Imagine receiving a paper from a student that incorporates the 3 mini-papers below as its main claims: What score would such paper earn?

Request: Write a paper with citations on why increased screen time affects sleep.

The impact of increased screen time on sleep has become a growing concern in recent years, as the use of technology continues to rise. Studies have shown that prolonged screen time can lead to a variety of sleep-related problems, including difficulty falling asleep, poor sleep quality, and shorter sleep duration. This paper will review the current research on the relationship between screen time and sleep, and discuss the underlying mechanisms that contribute to the negative effects of increased screen time on sleep.

One of the key factors that contribute to the relationship between screen time and sleep is the exposure to blue light. Blue light, which is emitted by electronic screens, suppresses the production of melatonin, a hormone that regulates the body's sleep-wake cycle (1, 2). As a result, exposure to blue light in the evening can disrupt the body's natural sleep-wake rhythm, making it harder to fall asleep and stay asleep (3, 4).

Additionally, the cognitive and emotional stimulation associated with screen use can also contribute to sleep problems. For example, using screens to engage in activities such as social media, gaming, and video streaming can activate the body's sympathetic nervous system and increase physiological arousal, making it harder to relax and fall asleep (5,6). Furthermore, studies have shown that using screens to consume emotionally stimulating content, such as news or social media, can also lead to increased feelings of stress and anxiety, which can further disrupt sleep (7,8).

Moreover, studies have suggested that the mere presence of screens in the bedroom, regardless of usage, may also affect sleep quality. (9) Bedroom screens such as televisions, computers, and smartphones can act as cues for wakefulness and discourage the development of a regular bedtime routine, and this in turn can disrupt sleep.

To conclude, the research suggests that increased screen time can have a significant impact on sleep. The negative effects of screen time on sleep are largely mediated by the suppression of melatonin production, as well as the cognitive and emotional stimulation associated with screen use. Individuals who are experiencing sleep problems should consider limiting their screen time in the evening, as well as avoiding screens in the bedroom.

References:

  1. Burkhart K, Phelps JR. Melatonin, sleep, and circadian rhythms: a complex relationship. Sleep Med Rev. 2009 Oct;13(5):307-20.
  2. Gooley JJ, Chamberlain K, Smith KA, Khalsa SB, Rajaratnam SM, Van Reen E, et al. Exposure to room light before sleep alters mood and cognitive function. Nature. 2011 Mar 17;472(7341):1209-12.
  3. Lockley SW, Cronin JW, Evans EE, Cade BE, Lee C, Landrigan CP, et al. Effect of reducing interns' weekly work hours on sleep and attentional failures. N Engl J Med. 2004 May 27;350(22):2134-41.
  4. Rajaratnam SM, Dijk DJ. Melatonin and the circadian regulation of sleep in humans. Sleep Med Rev. 2011 Oct;15(6):365-81.
  5. Chan JL, Lo EW, Choi KC, Yeung WF. A review of the effects of internet and video game addiction on sleep problems. Sleep Med Rev. 2013 Apr;17(2):115-23.
  6. Kuss DJ, Griffiths MD. Social networking sites and addiction: ten lessons learned. Int J Environ Res Public Health. 2011 Jul;8(7):3528-52.
  7. Bauer LA, Nelson E, Neufeld RW, Barriault C, Thomas KN. News exposure, perceived stress, and cortisol: a field study. Unfinished references section: 9 sources were cited in the paper, but most of these sources do not exist.

Request: Write a paper, with citations, on how much does reducing exposure to blue light improve sleep quality and amount.

The effects of blue light exposure on sleep quality and quantity have been studied extensively in recent years. Blue light, which is a type of light in the short-wavelength, high-energy visible spectrum, is known to suppress the production of melatonin, a hormone that regulates the sleep-wake cycle. As a result, exposure to blue light in the evening can make it harder to fall asleep and can lead to poorer sleep quality.

Several studies have investigated the impact of reducing blue light exposure on sleep. A study by Chellappa et al. (2018) found that exposure to blue-light blocking glasses for two hours before bedtime resulted in a significant increase in the production of melatonin and a significant improvement in sleep quality in a sample of healthy adults. The study found that the participants fell asleep faster, had a deeper sleep and felt more refreshed in the morning after wearing the glasses.

Another study by Burkhart and Phelps (2009) found that exposure to a device emitting blue light in the evening resulted in a significant reduction in sleepiness, an increase in cognitive performance, and a delay in the onset of melatonin production, as compared to exposure to a yellow light-emitting device. The study suggest that exposure to blue light before sleep have negative effect on sleep quality and delay the time that the melatonin starts to be produced.

In addition, A recent meta-analysis by Chang et al. (2020) found that exposure to blue light before bedtime is significantly associated with a reduced amount of sleep and poor sleep quality in both children and adults. The meta-analysis included a total of 21 studies and involved a total of 1215 participants, and suggest that blue light from electronic devices and energy-efficient lighting are having an negative impact on the sleep quality and duration.

However, It's also important to note that the effects of blue light exposure on sleep may vary depending on individual differences, such as age, sex, and pre-existing medical conditions.

In conclusion, several studies have found that reducing exposure to blue light in the evening can improve sleep quality and quantity. This can be achieved by using blue-light blocking glasses, avoiding electronic devices in the evening, or using dim red lights in the bedroom. However, more research is needed to fully understand the impact of blue light exposure on sleep, especially given the increasing use of electronic devices and energy-efficient lighting.

References:

Request: Write a paper with references on the relationship between videoconferencing and stress.

The use of videoconferencing technology has become increasingly common in recent years as a way to facilitate communication and collaboration among individuals and teams in a remote or geographically dispersed environment. However, while videoconferencing can offer many benefits, such as increased flexibility and productivity, it can also create additional stress for individuals, both in the short and long term.

The stress associated with videoconferencing can stem from a number of different sources, including the increased demands on individuals to appear presentable and engaged, the lack of nonverbal cues that are present in face-to-face interactions, and the added pressure to multitask during the call. Additionally, video conferencing can also exacerbate feelings of isolation and disconnection that are often experienced by individuals who work remotely.

One study found that videoconferencing can lead to increased feelings of social presence, which is the sense that one is interacting with real people and not just technology. This can result in higher levels of self-awareness, which can be stress-inducing, particularly for introverted individuals (Kirschner, & Kester, 2011). Another study showed that people experience greater mental effort and less social presence during videoconferences than in face-to-face interactions, and that this difference in experience can lead to increased stress levels (Chen, et al., 2020).

Furthermore, since videoconferencing is a form of “telepresence,” it can cause fatigue and other negative effects on mental and physical health. The study found that telepresence can lead to decreased productivity, and that people who frequently use videoconferencing technology to communicate with others may experience more feelings of stress and burnout as a result (Danaher, & Kennedy, 2018).

However, it's also important to note that video conferencing has been found to have positive effects on certain aspects of mental health. For example, one study found that, videoconferencing can have a beneficial effect on mental health outcomes such as reduced depression, anxiety and stress. (Galbraith, et al., 2020)

Overall, the relationship between videoconferencing and stress is complex, and the effects may vary depending on the individual and the context in which the videoconferencing is taking place. In order to minimize stress associated with videoconferencing, it may be helpful to establish clear expectations and guidelines for the use of the technology, to provide regular breaks, and to incorporate methods for maintaining social connections, such as virtual social events or team-building activities.

References:

Next, I thought to explore whether asking the same question again will produce a similar response. Thus, I repeated the question above. I was surprised to discover that while the argument remained the same – that videoconferencing has the potential to increase stress – the reasons why were somewhat different from the ones produced in the first response. Thus, an adept student should be able to pick the explanations that make the most sense, or that seem to have stronger support, from the two versions of the response, creating a stronger paper.

Request: Write a paper with citations on the relationship between videoconferencing and stress.

The widespread adoption of videoconferencing technology has revolutionized the way we work and communicate, allowing us to connect with colleagues, friends, and family members from virtually anywhere. However, despite its many benefits, the use of videoconferencing has also been linked to increased stress levels.

One study by Jia et al. (2021) found that participants who frequently used videoconferencing reported higher levels of stress compared to those who used it less frequently. This is thought to be due to the added social pressure and self-consciousness that can come with being on camera. Furthermore, the study found that participants who used videoconferencing for work-related purposes reported even higher levels of stress than those who used it for personal or leisure activities.

Another study by Rosen et al. (2020) found that participants who used videoconferencing for work reported higher levels of work-related stress, burnout, and emotional exhaustion. The study suggested that the added visual and auditory cues provided by videoconferencing can increase cognitive load, making it more difficult to multitask and manage competing demands.

Other research has also found that videoconferencing can lead to feelings of isolation and disconnection, particularly for remote workers (Kraut et al., 1998). This can be due to the lack of nonverbal cues and spontaneous interactions that are present in face-to-face conversations. Furthermore, studies have also suggested that people who use videoconferencing may feel more self-conscious about their appearance and environment, which can further contribute to stress (Meyer, 2009).

It's worth mentioning that studies are only showing correlation and not causality, further research need to be conducted to establish this relationship.

Despite these potential downsides, it is important to note that videoconferencing can also have a positive impact on stress levels. For example, videoconferencing can provide a sense of connection and support for remote workers, and can be an effective way to manage stress through social support (Thorsteinson et al., 2015). Additionally, many people find videoconferencing to be more convenient and less stressful than traveling for in-person meetings.

In conclusion, research has shown that videoconferencing can be associated with increased stress levels, particularly in the workplace. However, further research need to be conducted to understand causality. The increased pressure and self-consciousness caused by being on camera, added cognitive load, as well as isolation and disconnection can be contributing factors.

References

Jia, J., Huang, X., & Wang, D. (2021). The effect of videoconferencing on stress: A study of urban white-collar workers in China. Stress and Health.

Rosen, L. D., Whalon, K., & Salmon, J. M. (2020). The Impact of Videoconferencing on Stress, Burnout, and Emotional Exhaustion. Journal of Applied Psychology, 105(6), 717–732. https://doi.org/10.1037/apl0000588

Kraut, R., Kiesler, S., Boneva, B., Cummings, J., Helgeson, V., & Crawford, A. (1998). Internet Paradox: A Social Technology that Reduces Social Involvement and Psychological Well-being? American Psychologist, 53(9), 1017–1031. https://doi.org/10.1037/0003-066X.53.9.1017

Meyer, J. P. (2009). How Videoconferencing Affects Social Inter   The chatbot yet again did not finish the references section, and only the last source is authentic

So, in spite of drawing on mostly made-up sources, the narrative produced by the chatbot is clear and coherent. Granted, not all students might be able to articulate the questions with sufficient precision or work out the logic behind asking the chatbot to expand on a portion of its response in a separate response. Yet with some trial and error, students will learn how ChatGPT works and how to both ask questions in a manner that results in a response that yields the expected results and combine the various responses into a cohesive research paper.

And even if the result students get may be lacking in depth, breadth, or analysis, I suspect many will agree that ChatGPT’s writing is clear and evidence-based, and the acknowledgement of limitations adds a critical reasoning component to an otherwise unimaginative literature review. In my experience, it takes most students considerable time to develop these skills. Now, the question in my mind is what I as the instructor can do to acknowledge and utilize this new technology in the college classroom.

Next, I focused my efforts on writing tasks that provide scaffolding for skills required to produce a research paper.

Goal 2: Smaller tasks that scaffold the skills to write a research paper

  1. Generation of ideas for a paper on a given topic.
  2. Explaining a research finding.
  3. Making writing more concise.
  4. Asking for a bibliography on a particular topic.
  5. Paraphrasing and summarizing an abstract of a research study.
  6. Defining/explaining a term.
  7. Integrating multiple research abstracts into a literature synthesis.

1. Request: I'd like to design a novel research study to investigate the relationship between videoconferencing and sleep. What unanswered questions exist on this topic?

There are a number of unanswered questions related to the relationship between videoconferencing and sleep. Here are a few that come to mind:

  1. How does the use of videoconferencing for work or school impact sleep quality and quantity for adults and children?
  2. Is there a difference in the impact of videoconferencing on sleep for people who use it occasionally versus those who use it frequently or daily?
  3. Does the type of videoconferencing platform (e.g. Zoom, Google Meet, etc.) have an impact on sleep?
  4. What are the effects of late-night videoconferencing on melatonin levels and circadian rhythms?
  5. How do different lighting conditions during videoconferencing (e.g. natural light vs. artificial light) affect sleep?
  6. Does prolonged use of videoconferencing contribute to symptoms of virtual fatigue or "Zoom fatigue"?
  7. Can interventions such as adjusting lighting or taking regular breaks during videoconferencing improve sleep quality?
  8. How does the duration of video conferencing during the day affect sleep patterns?
  9. How does the type of work affect the sleep patterns of employees on video conferencing
  10. How does the satisfaction with work-life balance, social support and sense of belonging to an organization affect the sleep of employees who use videoconferencing for their work?

These are just a few of the questions that could be explored in a research study on the relationship between videoconferencing and sleep. Keep in mind that a good research study will be well-designed, with clear objectives, hypotheses, and methods that allow you to effectively address the questions you're interested in investigating.

I would recommend you also to conduct a comprehensive literature review, which will allow you to identify the existing knowledge and what are the gaps that need to be filled.

The bot's response is extensive and presents a number of interesting ideas. For me, ideas #1, #4, and #6 in particular have potential as research questions. In class assignments, I often ask students to come up with novel research ideas or interesting topics, and it is typically a challenging and time-consuming task for all students. For the chatbot, it only took 15 seconds to generate these ideas.

Moving on, let’s say that I notice a pattern in the data I’ve collected, but I either don’t have a good explanation for it or I want to brainstorm explanations. Chatbot to the rescue! I present 2 versions of the response obtained on different days. In the first one, I neglected to ask for citations, so the output does not cite sources. In the second one, I asked for citations explicitly.

2A. Request: Why might personal videocalls improve sleep, whereas work-related videocalls impair sleep?

(here, I neglected to ask to include sources, so the output was produced without citations)

There is research that suggests that video calls, particularly those that are work-related, can have a negative impact on sleep. This may be due to a number of factors, such as the stress of feeling the need to be "on" during the call, the pressure of having to perform well, and the potential for the call to run longer than expected. Additionally, work-related video calls may occur at times when an individual would normally be winding down for the evening, and the bright screens and artificial light associated with video calls can also disrupt the body's natural circadian rhythm and make it harder to fall asleep.

On the other hand, personal video calls may have a positive impact on sleep, as they can help to foster feelings of connection and social support, which are important for overall well-being. Additionally, since personal video calls are likely to occur during times when an individual is already awake, they may not disrupt the body's natural sleep-wake cycle as much.

It's also worth noting that the overall impact of video calls on sleep can also vary depending on an individual's specific circumstances and preferences. For example, someone who finds video calls to be draining may benefit from scheduling them earlier in the day, or taking breaks in between calls to rest and recharge. Additionally, practices like turning off notifications and dimming the screens of electronic devices before bed, can also help reduce the negative impact of video calls on sleep.

It's also good to keep in mind that other factors that can impact sleep too, like exercise, stress levels, diet and overall good sleep habits should be part of a good routine to maintain healthy sleep quality.

2B. Request: Why might personal videocalls improve sleep, whereas work-related videocalls impair sleep?

(here, I asked for citations explicitly, but note that the sources provided by the chatbot are made up.)

Research has suggested that personal videocalls may improve sleep, whereas work-related videocalls may impair sleep.

A study published in the Journal of Sleep Research in 2020, "The Impact of Videoconferencing on Sleep: A Review of the Literature" by J.M. Colrain, M.K.A. Roane, and J.A. Baker, found that personal videocalls were positively associated with sleep, as they were reported to improve social connections and reduce feelings of isolation. This study also reported that work-related videoconferencing was negatively associated with sleep, as it was reported to increase stress and workload.

Another study published in Journal of Behavioral Addictions in 2018, "The impact of self-reported technology use on sleep patterns and sleep disorders among college students" by M.J. Kostygina, T.F. Harvey, J.L. Hale, and T.L. Rosenberg, showed that self-reported social technology use was positively associated with sleep quality and duration, while self-reported work-related technology use was negatively associated with sleep quality.

It's worth noting that these studies are old and that more recent research is needed to fully understand the relationship between videoconferencing and sleep. Additionally, further research is needed to understand if these results are consistent across individuals, cultures and demographics. It's also important to keep in mind that every individual have their own sleep needs and habits, so the impact of video conferencing on sleep will vary on person to person.

Please note that this information may not be up to date, if you want to find the most recent research studies you can try searching for papers on academic databases such as PubMed or Google Scholar.

While this response is not sufficient for any kind of professional work, it provides plenty of leads to pursue additional research. For a student paper, however, the response may be sufficient to explain a particular pattern in the data... even if the sources that ostensibly support the explanation are bogus.

3A. You can also ask the chatbot to make either its response, or a paragraph of text, more concise.

Below, I requested that it condense the previous response on the contrasting effects of work-related vs. personal video calls on sleep (see above).

Research has suggested that personal videocalls may improve sleep, whereas work-related videocalls may impair sleep. Studies have found that personal videocalls can reduce feelings of isolation and improve social connections, while work-related videocalls have been found to increase stress and workload. It's worth noting that more recent research is needed to fully understand the relationship between videoconferencing and sleep, as well as understanding the impact on individuals, cultures, and demographics

3B. This is what happens when I place a paragraph from a student paper into the ChatGPT and asking it to make it more concise.

Request: Make this more concise: Studies conducted by Maria Clark in 2020 shows the drastic correlation of social media increase and mental health decrease. In 2011, only 35% of United States citizens had social media, and that number spiked to 81% by the end of 2019. As these statistics increased, so did the negatives. Between 2012 and 2015, cases of depression diagnosed in females increased by 50%, with 21% in males. In grades 8-12, depressive symptoms grew by 33%, and the suicide rate during this time increased by an astonishing 65%. In total, by 2019, the suicide and self-harm rate in individuals under 18 was up by about 150% from what it was in 2015. All these problems had social media as a direct factor. Those who spend over 3-5 hours daily and over 10-15 hours a week on social media are 56% more likely to have lower mental health levels and exhibit more symptoms of depression than an individual who spends less than an hour a day on social media.

A 2020 study by Maria Clark found a strong correlation between increased social media use and decreased mental health. In 2011, 35% of US citizens had social media, rising to 81% by 2019. This increase correlated with a rise in depression, with cases increasing by 50% in females and 21% in males between 2012 and 2015. Additionally, depressive symptoms among students in grades 8-12 rose by 33%, and suicide rates increased by 65%. By 2019, suicide and self-harm rates among individuals under 18 were 150% higher than in 2015. This study suggests that individuals who spend over 3-5 hours daily and over 10-15 hours a week on social media are 56% more likely to have lower mental health levels and exhibit more symptoms of depression than those who spend less than an hour a day on social media.

Thus, the chatbot can make writing more concise.

4. Asking for a bibliography. Here, let’s say I’m interested in doing a research study on a particular topic but prefer not to comb the literature to figure out what some of the unanswered questions/gaps in knowledge are. Why not ask the Chatbot to do it for me? Yes, most of these sources are made up, and the explanation from the chatbot about its limitations misleads one to view them as authentic, but the output looks impressive nonetheless, particularly to someone not familiar with the work in the field.

I apologize, but I am unable to provide a list of recent research studies on the relationship between videoconferencing and sleep, as I don't have access to current databases, my last knowledge cut off was 2021 and I don't have access to the internet. However, here are some examples of studies that have been conducted on the topic:

  1. "The impact of videoconferencing on sleep: a review of the literature" published in Journal of Sleep Research in 2020 by J.M. Colrain, M.K.A. Roane, and J.A. Baker.
  2. "The effect of technology use on sleep: a systematic review" published in Journal of Clinical Sleep Medicine in 2016 by M.A. Gradisar, J.L. Wolfson, T.F. Harvey, J.L. Hale, and T.L. Rosenberg.
  3. "The impact of videoconferencing on sleep and circadian rhythm" published in Journal of Biological Rhythms in 2017 by J.A. Baker and M.K.A. Roane.
  4. "The effect of electronic devices on sleep patterns and sleep disturbances" published in Journal of Sleep Research in 2015 by M.A. Gradisar, T.F. Harvey, J.L. Hale, and T.L. Rosenberg.
  5. "The effect of light from electronic devices on the melatonin levels of adolescents" published in Journal of Pineal Research in 2013 by C.P. Chellappa, S.A. Steiner, D.S. Sanders, and D.S. Provencio.
  6. "The impact of evening use of light-emitting eReaders on sleep, circadian timing, and next-morning alertness" published in Proceedings of the National Academy of Sciences in 2014 by J.A.C. Charles, J.L. Cheung, A.C.T. Philipps, and J.A. Czeisler.
  7. "The impact of screen time on sleep in children and adolescents: a systematic literature review" published in Sleep Medicine Reviews in 2019 by E.N. Van der Lely, R.A.C. Van der Stelt, and E.A. Van Someren.
  8. "The effects of blue-light exposure on circadian rhythms, sleep and cognitive function in adolescents" published in Journal of Sleep Research in 2020 by L.S. Graveling, N.E. Fogg, and J.F. Duffy
  9. "The impact of self-reported technology use on sleep patterns and sleep disorders among college students" published in Journal of Behavioral Addictions in 2018 by M.J. Kostygina, T.F. Harvey, J.L. Hale, and T.L. Rosenberg.
  10. "The Impact of Mobile Technology Use on Sleep Quality and Quantity in Children and Adolescents: A Review of the Literature" published in Journal of Child Neurology in 2020 by B.C. Dreyer and J.J. Kaczmarski.

Please note that these studies are from before my knowledge cut off and some of them may not have direct relationship with videoconferencing. Also, some may no longer be available, or may have been updated after 2021. This list is just to provide you with an idea of the type of studies that have been conducted on the topic.

5. Paraphrasing and summarizing. I asked the chatbot to paraphrase and then summarize an abstract of a research study accessible here: https://psycnet.apa.org/record/2015-46247-001)

5A. Request: Can you paraphrase the following paragraph: Our attention and memory can be biased toward objects having high self-relevance, such as things we own. Yet in explaining such effects, theorizing has been limited to psychological determinants of self-relevance. Here we examined the contribution physical actions make to this ownership bias. In Experiment 1, participants moved object images on a touch interactive table that either arbitrarily belonged to “self” or “other,” and that were moved into locations closer or farther from their bodies. Subsequent recognition was highest for self-owned objects moved closer to the body, as measured via a subsequent memory recall test. In Experiment 2, when participants moved images via keyboard rather than overt action, the proximity effect of the body on attention was abolished. In Experiment 3, participants pulled or pushed self-owned or other-owned object images to side-by-side locations on a touch interactive table. Self-owned objects that were pulled were recognized the most. Our findings demonstrate that physical actions can have a direct impact on the psychological saliency of owned objects, with the act of bringing objects toward the self leading to greater recall.

The study discusses how our attention and memory can be biased towards objects that are deemed as self-relevant, such as things we own. The study examines how physical actions contribute to this bias in memory recall. In the first experiment, participants interacted with object images on a touch screen, moving them closer or farther from their bodies, and subsequently recall was highest for self-owned objects that were moved closer to the body. The second experiment found that when the images were moved via keyboard rather than physical action, the proximity effect on attention was removed. The third experiment found that participants had the best recall of self-owned objects that were pulled closer to themselves. Overall, the study suggests that physical actions play a direct role in how we remember objects that are self-relevant.

When running the 2 versions of the text – the original abstract and the paraphrase – through textcompare.org, I learned that the paraphrasing ended up being too close to the original text, so I would mark it as plagiarism in student work (in addition to copying entire phrases verbatim, the idea units were not re-ordered).

The output below is from textcompare.org. Text in black = same in both version; Green = only appears in ChatGPT’s paraphrase. Red = only appears in the original text.

OThe study discusses how our attention and memory can be biased towards objects thaving hight are deemed as self-relevancet, such as things we own. Yet in explaining such effects, theorizing has been limited to psychological determinants of self-relevance. Here we examined the contributionThe study examines how physical actions makcontribute to this ownership bias. In Ebias in memory recall. In the first experiment 1, participants movedinteracted with object images on a touch interactive table that either arbitrarily belonged to “self” or “other,” and that were moved into locationsscreen, moving them closer or farther from their bodies. S, and subsequently recognitionall was highest for self-owned objects that were moved closer to the body, as measured via a subsequent memory recall test. In Experiment 2, when participants moved images. The second experiment found that when the images were moved via keyboard rather than overtphysical action, the proximity effect of the body on attention was abolished. In Experiment 3, participants pulled or pushed self-owned or other-owned object images to side-byside locations on a touch interactive table. Self-owned objects that were pulled were recognized the most. Our findings demonstrateremoved. The third experiment found that participants had the best recall of self-owned objects that were pulled closer to themselves. Overall, the study suggests that physical actions can haveplay a direct impact on the psychological saliency of owned objects, with the act of bringing objects toward the self leading to greater recallrole in how we remember objects that are self-relevant. Removed: 17

Added: 19

Generated at https://www.textcompare.org/es/ on 1/10/2023, 3:35:11 PM

However, the chatbot does rather well summarizing an abstract of a research study:

5B request: Provide a summary of the following paragraph: Our attention and memory can be biased toward objects having high self-relevance, such as things we own. Yet in explaining such effects, theorizing has been limited to psychological determinants of self-relevance. Here we examined the contribution physical actions make to this ownership bias. In Experiment 1, participants moved object images on a touch interactive table that either arbitrarily belonged to “self” or “other,” and that were moved into locations closer or farther from their bodies. Subsequent recognition was highest for self-owned objects moved closer to the body, as measured via a subsequent memory recall test. In Experiment 2, when participants moved images via keyboard rather than overt action, the proximity effect of the body on attention was abolished. In Experiment 3, participants pulled or pushed self-owned or other-owned object images to side-by-side locations on a touch interactive table. Self-owned objects that were pulled were recognized the most. Our findings demonstrate that physical actions can have a direct impact on the psychological saliency of owned objects, with the act of bringing objects toward the self leading to greater recall.

This study examines how physical actions contribute to the bias in attention and memory towards self-relevant objects, such as things we own. The study conducts 3 experiments, in the first experiment, participants moved object images on a touch screen, and recall was highest for self-owned objects moved closer to the body. The second experiment, found that when images were moved via keyboard the proximity effect on attention was removed. In the third experiment, recall was better for self-owned objects pulled closer. The study concludes that physical actions play a direct role in memory recall of self-relevant objects.

I ran the 2 versions of the text through textcompare. The summary is a lot better than the paraphrase in avoiding plagiarism. If I were reading this summary as a submission from a student, I would consider it of good quality.

Our attention and memory can be biased toward objects having highThis study examines how physical actions contribute to the bias in attention and memory towards self-relevancet objects, such as things we own. Yet in explaining such effects, theorizing has been limited to psychological determinants of self-relevance. Here we examined the contribution physical actions make to this ownership bias. In EThe study conducts 3 experiments, in the first experiment 1, participants moved object images on a touch interactive table that either arbitrarily belonged to “self” or “other,” and that were moved into locations closer or farther from their bodies. Subsequent recognition was highest for self-owned objects moved closer to the body, as measured via a subsequent memory recall test. In Experiment 2, when participants moved images via keyboard rather than overt action,screen, and recall was highest for self-owned objects moved closer to the body. The second experiment, found that when images were moved via keyboard the proximity effect of the body on attention was abolished. In Experiment 3, participants pulled or pushed self-owned or other-owned object images to side-by-side locations on a touch interactive table. Self-owned objects that were pulled were recognized the most. Our findings demonstrateremoved. In the third experiment, recall was better for self-owned objects pulled closer. The study concludes that physical actions can haveplay a direct impact on the psychological saliency of owned objects, with the act of bringing objects toward the self leading to greater recallrole in memory recall of self-relevant objects. Removed: 9

Added: 7

Generated at https://www.textcompare.org/es/ on 1/10/2023, 3:38:12 PM

6. Defining/explaining a term. Here, I asked the chatbot to define a well-known phenomenon in psychology where people prefer the letters in their name to other letters of the alphabet.

Request: What is a name-letter effect?

The name-letter effect refers to the phenomenon where people have a tendency to prefer the letters in their own name over other letters. This effect has been observed in a variety of studies, in which people have been found to have a positive attitude towards letters that appear in their own name. For example, in one study, participants were asked to rate the desirability of different letters, and those participants whose names contained a particular letter rated that letter more favorably than those whose names did not contain that letter.

The effect is considered as a minor form of self-esteem enhancement, people tend to like things that are related to themselves and thus, because letters of our name are directly related to oneself people tends to like them more, this way the name-letter effect has been found to be related with various self-esteem, identity, and personality measures.

It's also important to notice that this effect can occur in different languages and cultures, and it is not limited to individuals' first names, but also applies to their last names, nicknames, and even to the initials of their names.

The first sentence of the chatbot’s response shares considerable commonality with Internet sources such as Wikipedia, although in all fairness there is a finite ways in which you could define the name-letter effect. I could not tell (via a google search of various sentences and phrases) whether the rest of the response shares phrasing with other sources.

7. Finally, I pasted citations and abstracts of 3 research studies into the request textbox and asked ChatGPT to 'Create a single paragraph that integrates the findings of the following abstracts of 3 research studies.' While I am not including the request here because it was lengthy, the output produced by the chatbot was impressive (feel free to try it with several articles in your field of study). The chatbot began with a topic sentence that captured the take-away from the articles, followed by a one- or two-sentence summary (with statistics!) of at least one key aspect of each study (all 3 aligned with the take-away), and concluded by restating the take-away. If students are able to locate 3-4 abstracts of relevant research studies and feed them into the chatbot, the paper produced by ChatGPT will be fluent, on point, and include authentic citations.

Conclusions

To conclude, OpenAI’s newest chatbot, ChatGPT, can be a powerful tool when it comes to academic writing. It is capable both of performing tasks foundational to good writing in scientific fields, such as summarizing sources, revising writing for concise expression, and explaining a concept, and of tasks that incorporate these skills into a longer narrative, such as a synthesis of several sources or a research paper. ChatGPT also seems to be able to integrate evidence from multiple sources and from opposing points of view and offer basic analysis. For a student learning scientific writing, these skills are challenging to acquire and are typically mastered over multiple semesters or even years of college.

The chatbot can also be used to generate thesis ideas and potential research directions, although it is limited by the database from which it draws information, and it is unclear how novel those ideas might be (although to a student without expertise in a given domain, many ideas generated by ChatGPT may seem powerful and new). Interestingly, in my requests, the evidence ChatGPT cited in its response appeared to be scholarly without me having to specify that it be so. Herein lies a caution, however:

The current version of ChatGPT is subject to several limitations. One is its inability to align cited evidence with sources. Most of the citations provided by ChatGPT are made up, even though they look legitimate. It also lacks depth when asked to generate an evidence-based argument or integrate scholarly literature, so its output works very well as an outline but not so well as a finished product. To overcome the first limitation, one could feed the chatbot the text of actual articles or abstracts; to overcome the second limitation, one could ask the chatbot to produce a summary of a particular article or abstract of an article. Neither obstacle is insurmountable to a determined user, particularly to someone willing to use trial and error to achieve the desired result. Questions remain, however, about whether or not ChatGPT is capable of moving beyond summaries and rather basic synthesis of evidence to generate a sophisticated and factually accurate argument that reflects what most of us aspire to when teaching college-level writing.

I am left pondering my next task: How to acknowledge the capabilities of artificial intelligence in the classroom, how to educate the students on its strengths and limitations, and on the most basic level, how to approach the use of chatbots like ChatGPT by students in and out of class.

I include a statement on my syllabus that a student may not turn in work completed by someone else as their own without acknowledging the source, and that failing to do so will constitute academic dishonesty. Yet, given that it is currently difficult or impossible to determine whether or not students use a chatbot in their writing (but see https://www.npr.org/2023/01/09/1147549845/gptzero-ai-chatgpt-edward-tian-plagiarism), and that the chatbot’s output is produced using natural language and follows writing convention (at least in the examples I’ve provided), we may not be able to detect when a student uses writing produced by artificial intelligence in their submissions. Moreover, most of us already use artificial intelligence to improve our writing (for instance, by relying on Microsoft Word’s grammar and usage suggestions or on the auto-complete when composing a text), although the capabilities of ChatGPT go well beyond what we are used to.

Perhaps we should be open about this technology with our students and discuss what it can and cannot do for us? Perhaps we should put our writing prompts through the chatbot ahead of time and discuss the results in the classroom or create assignments that will be challenging for a chatbot to crack? I don’t have the answers to these questions, but I suspect the next few weeks and months will bring them to the forefront of academic discussion.