Next Article in Journal
Dietary (Poly)phenols in the Management of Chronic Kidney Disease: A Narrative Review
Previous Article in Journal
Bariatric–Metabolic Surgery: The State of the Art and the Management of Complications
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Dietetics
Volume 4
Issue 4
10.3390/dietetics4040050
dietetics-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

AI Recipe Blog Is Evaluated Similarly to a Recipe Blog Created by Nutrition and Dietetic Students

by
Katie N. Kraus
*,
Stacy L. Bevan
,
Sarah Moore Smith
,
Maeci H. Armstrong
,
Brooke Campbell Jeppesen
,
Catherine Fish
and
Heidi J. Wengreen
Department of Nutrition, Dietetics & Food Sciences, Quinney College of Agriculture and Natural Resources, Utah State University, Logan, UT 84322, USA
*
Author to whom correspondence should be addressed.
Dietetics 2025, 4(4), 50; https://doi.org/10.3390/dietetics4040050
Submission received: 28 August 2025 / Revised: 23 September 2025 / Accepted: 9 October 2025 / Published: 1 November 2025
Browse Figure
Versions Notes

Abstract

With the growing use of AI, it is important to know target audiences’ perceptions of its use. A convenience sample of students were invited to take an online survey in which they were randomly assigned to Group 1 (evaluated a student-generated blog; n = 456) or Group 2 (evaluated an AI-generated blog; n = 492). The results of independent t-tests and chi-squared tests indicated no group differences in ratings of ease of recipe preparation, time to prepare the recipe, utilization of common ingredients, and frequency of intended use of the blog. The student-generated blog was rated higher on budget friendliness (p = 0.025). A total of 42% indicated they would be less willing to use a blog if they knew it was AI-generated, while 43% indicated that it would make no difference and 4.4% indicated being more likely to view the AI-generated blog. Two researchers used a thematic analysis approach to evaluate participants’ free responses regarding the likelihood of using a recipe blog that was AI-generated. Participant perceptions of an AI-generated blog ranged from very positive to very negative. Some themes highlighted the potential benefits of AI or a more neutral stance indicating that “a recipe is a recipe”. The majority of themes highlighted the benefits of content that was created, verified, or tested by humans, or espoused a human touch. Students should be trained to cater to consumer preferences, and to add value in a world that includes AI-generated content.

1. Introduction

The use of artificial intelligence (AI) has expanded significantly over the past few years in many areas of life, with education being no exception [1]. A meta-analysis found that AI use among higher education students improved cognitive, technical, and interpersonal skills [1]. Yet Msambwa et al. cautioned that integrating AI into the classroom should be enacted strategically, with policies to maintain academic integrity and a complementary balance of AI to human efforts in education [2].
The promising benefits of AI in education have been met with reservations about ethical use, academic integrity, concerns of overreliance resulting in a loss of critical thinking and communication skills, and fears of teachers being replaced with AI [1,3].
In the field of dietetics, AI tools like ChatGPT have an abundant potential to assist Registered Dietitian Nutritionists (RDNs) in practice. AI is beginning to be utilized in meal planning and recipe-related tasks [4]. For example, a 2023 study reported that an AI-powered meal planner created customized meal plans that were accepted by the end user [5]. Chatelan et al. described that ChatGPT successfully created 1-day menus and accompanying recipes appropriate for individuals with type 2 diabetes and individuals undergoing hemodialysis [4]. However, the authors note that ChatGPT responses vary over time, even when the same prompt is used [4]. Researchers also report that AI is not infallible and has included foods inappropriate for the specified dietary restrictions when making meal plans [4,6]. YouTube and Tiktok chefs have been challenging human-generated recipes against AI-generated recipes for quality and taste, with the chefs’ recipes generally winning [7]. Professionals should be aware that AI should not replace their expertise and judgments, as it occasionally provides inaccurate or biased responses [3,4]. AI generation should always be reviewed and evaluated by professionals prior to dissemination.
With the growing availability of generative AI tools, it is increasingly important to understand how AI-generated content is perceived by target audiences. Understanding these perceptions has implications for both digital health communication practice and the integration of AI tools in nutrition and dietetics education.
In addition to recipe development, AI technology is now being used to develop social media content. AI-generated text has been detected on social media platforms like Medium, Quora, and Reddit, with an increase on some platforms since the release of ChatGPT 3.5 [8]. According to a National Public Radio article, some food bloggers are striving to distinguish their blogs from AI-generated content, while others are embracing AI technology [7]. Zhou et al. warned that AI-generated text could spread misinformation, negatively impacting the public [9]. Previous research has not assessed the use of AI in creating a recipe blog or end users’ perceptions of a blog that features AI-generated content [9].
Given the high percentage of individuals turning to food blogs, and their potential for disseminating food and nutrition information, dietetic curriculum should educate students on creating science-based and engaging content for social media platforms such as blogs [10], while addressing the ethical use of AI tools. Incorporating blog-based assignments into nutrition curricula supports the development of essential communication competencies, including the ability to convey evidence-based information in an accessible and engaging format. Budget-friendly, quick, and easy recipes that are readily accessible on recipe blogs are a timely priority for college students who are prone to food insecurity, with limited resources and time [11,12]. For example, students in a Food Literacy course at Utah State University (USU) create recipe blog posts that aim to address college students’ barriers to healthy eating by focusing on recipes that can be prepared quickly and easily and are both inexpensive and healthy. The purpose of this study was to determine the differences in college students’ ratings of a recipe blog created by students and a recipe blog that includes AI-generated content.

2. Materials and Methods

This cross-sectional comparative study examined the differences in student preference between a student-generated food blog and an AI-generated food blog. All procedures were reviewed and approved by the USU institutional review board (Protocol #13819); this study was classified as exempt. A convenience sample of USU students were primarily located in Logan, UT, but the survey was completed online. The study took place during October–December 2023.

2.1. Blog Creation

2.1.1. Student-Generated Blog

Students in a Food Literacy course created a blog post as a course assignment. The blog was created using WordPress and utilized the Dynamic Recipe Card feature. Posts included a personable commentary on the recipe, step-by-step instructions, cost analysis, and photos of the process and end result.

2.1.2. AI-Generated Blog

Researchers created a blog aimed at mimicking the student-generated blog made by students in the Food Literacy course. Researchers drafted a list of recipes that may be commonly consumed by college students. ChatGPT (GTP-4) was then used to draft the recipes for the predetermined recipe titles. The following prompts were used as needed to create blog posts that were similar to the student-generated blog posts:
  • “Give me a simple blog post for [insert recipe name here]”
  • “Provide more commentary on this post: [inserted recipe instructions]”
  • “Write a brief description for: [insert recipe name here]”
  • “What is the cost of these ingredients for an entire container [insert list of ingredients (without amount)]?”
Researchers also edited the AI-generated text as needed. For example, often the introductory information highlighted the history of a recipe. Because the history of the recipe was not included in the student-generated blog, this topic was omitted. If the generated introductions were too similar to each other (e.g., frequent use of wording such as “In the world of culinary delights…”), revisions were made to vary the wording. However, no changes were made to recipe content (ingredients, amounts, cost, cooking times, etc.). Recipes were not prepared or tested for accuracy prior to being included in the blog. Cover images and images that aligned with steps in the process of preparing the recipe (e.g., cutting tomatoes, cooking chicken, etc.) were generated using the AI image generator in Canva. The website layout and navigation of the AI-generated blog were formatted to mimic the student-generated blog that was created by students in the Food Literacy course.

2.2. Recruitment

Participants were recruited via fliers, social media, TV monitor announcements, informational tables on campus, and word of mouth. An email inviting individuals (students 18 years of age or older) to complete a brief, anonymous survey was sent to 17,879 USU students. Interested students reviewed a Letter of Information that described the purpose of the study, the length of the survey, incentive (voluntary entry into a drawing for 1 of 16 $25 Amazon gift cards), and the strategies used for safe data collection and storage. The Letter of Information ended with these statements: “By continuing to the online survey, you agree that you are 18 years of age or older and wish to participate. You agree that you understand the risks and benefits of participation and that you know what you are being asked to do. You also agree that you have contacted the research team with any questions about your participation, and are clear on how to stop your participation in this study if you choose to do so.”
The online Qualtrics survey randomly assigned participants to one of two groups: (1) participants who evaluated the student-generated recipe blog [Student-Generated Blog], or (2) participants who evaluated the blog with AI-generated content [AI-Generated Blog]. Participants were advised to review at least two recipes; participants were not advised to prepare any of the recipes prior to providing evaluative feedback on the respective recipe blog. Information about potential confounding factors was also gathered (e.g., confidence in cooking skills, prior enrollment in the Food Literacy course, age, gender, class rank, and major). After completing the survey, participants were made aware of which group they had been assigned to, and were provided a link to the other blog for their review if they were interested.

2.3. Survey

All participants completed a 26-question survey that gathered information about participant demographics, assessment of the blog on various criteria related to their perceptions of the recipes (budget friendliness, ease of preparation, time to prepare, utilization of common ingredients, how often the participants thought they would use the blog), the impact, if any, it would make if they knew beforehand that the recipe website was AI-generated, and any additional comments.

Survey Creation and Validation

Researchers drafted the survey. Cooking self-efficacy was assessed with three applicable statements from the Food Literacy Assessment Tool (FLitT) (see Table 1) [13], as well as a statement added by the researchers; these four questions had an acceptable internal consistency level (Cronbach’s Alpha = 0.664) [14]. The survey was designed to assess user ratings on several dimensions relevant to the practical use and credibility of information shared on the blog. These dimensions included budget friendliness, ease of preparation, time to prepare, use of common ingredients, and frequency of intended use. Student researchers then pre-tested the survey to evaluate the usability and the content of the survey. Three researchers who are RDNs also reviewed the survey content. Changes were made based on these evaluations. Overall, the student researchers and RDN researchers agreed that the survey accurately assessed perceptions of the recipe blogs, indicating face validity.

2.4. Analysis

2.4.1. Quantitative Data Analysis

Quantitative Data was analyzed using IBM Statistical Package for Social Sciences (SPSS, version 30). The predictor variable was the group (Student-Generated Blog vs. AI-Generated Blog). Outcome variables included participants’ ratings of the recipe blog in terms of budget friendliness, ease of preparation, time to prepare, utilization of common ingredients, and how often participants thought they would use the blog. Participants rated the criteria on a 5-point Likert scale ranging from Strongly Disagree (1) to Strongly Agree (5) (as well as “Don’t know” or “Prefer not to answer”). Group differences were assessed using independent t-tests and chi-squared distributions. Significance was set at p < 0.05.

2.4.2. Qualitative Data Analysis

Participants’ free responses to the prompt “Please provide any additional comments regarding your likelihood of using a recipe blog that was AI-generated” were analyzed by two researchers using a thematic analysis approach [15]. The two researchers separately read through all comments and made memos of any initial ideas or concepts. They reviewed the first third of the responses and identified possible codes. They then met to adjudicate any discrepancies in coding and established a common code book. The code book included all codes the researchers agreed upon, as well as definitions, inclusion criteria, exclusion criteria, and representative quotes. Separately, the researchers analyzed the next third of the responses and met again to discuss any additional codes that were added and to adjudicate discrepancies, as well as combine and refine codes as needed. This process was repeated for the remaining third of responses. A master spreadsheet was created that housed the agreed upon codes and quotes that aligned with each code. The researchers then discussed all codes, including patterns and themes that were observed. Trustworthiness was promoted by adopting a specific research method (thematic analysis [15]) and through regular debriefing between the two researchers [16]. Because participants’ free-response comments often did not align with their stance on their likelihood of using a recipe blog that was AI-generated, qualitative responses were analyzed and summarized together.

3. Results

3.1. Participants

The exact response rate cannot be determined because it is unknown how many students viewed fliers, etc. Based on email invitations alone, of the 17,879, 1536 participants accessed the survey; however, only 948 were included in the analysis (see Figure 1 for additional information), indicating an 8.6% response rate and a 5.3% eligible response rate. Of the 948 participants, 456 (48%) were randomly assigned to the Student-Generated Blog group and 492 (52%) were assigned to the AI-Generated Blog group. There were no group differences in age, gender, class rank, major, percent who were enrolled/had previously been enrolled in the Food Literacy course, or cooking confidence.

3.2. Assessment of the Recipe Blogs

As seen in Table 2, the student-generated blog was rated higher for its budget friendliness (p = 0.025). The majority of students indicated they would use the blog they reviewed 1–2 times per month or 1–2 times per week; there were no differences between the Student-Generated Blog group and the AI-Generated Blog group (see Table 2). Nearly half (n = 409, 43.11%) indicated that knowing that a recipe website was AI-generated would make no impact on how likely they would be to use the website, few (n = 37, 4.4%) indicated they would be more likely to use the website, 359 (42.8%) indicated that they would be less willing to use the website, 23 (2.7%) selected “other”, and 11 (1.3%) preferred not to answer. Those who selected “other” indicated being unsure or wanting additional information such as comparing the recipes to their own cooking knowledge or reading reviews of the recipes. When comparing the responses of the Student-Generated Blog group and the AI-Generated Blog group, there was no difference in likelihood of using the website if they knew it was AI-generated (p = 0.666). For this reason, the participants’ comments (n = 775) were analyzed together rather than separated by group. Similarly, comments were also not analyzed separately based on participants’ likelihood of using a recipe blog if they knew it was AI-generated (e.g., more likely vs. less likely vs. no difference) because it was not uncommon for participants’ comments to contradict their stated likelihood of using the recipe blog.
Participants’ perceptions of AI existed on a spectrum, as displayed in Table 3. Additional main themes and representative quotes are summarized in Table 4, organized generally from positive to negative.
Some participants recognized that the AI blog was AI-generated, while others could not tell the difference. One participant wrote, “When I was reading through the recipes, the description and manner of everything felt like AI. I didn’t want to say anything previously to belittle the creativeness of the phrases, but I definitely felt like it was written by ChatGPT. I now can confirm my suspicions to be true.” (N142), while others wrote “I was surprised to learn the website I saw was AI generated” (N130) and “If I wasn’t told, I wouldn’t know” (N383). Some comments highlighted the impact of previous experience with AI (general or recipe related) on their negative or positive perceptions of an AI-generated blog. Some comments indicated being impressed that the recipe blog was AI-generated, “I think that it’s cool AI was able to create this.” (N133), while others voiced concern that AI “is just not there yet” (L143). Some comments highlighted that students didn’t necessarily care if the blog was AI-generated, but preferred that the use of AI be disclosed. For example, one participant wrote, “Initially after hearing that the blog was AI-generated, I was a little uncomfortable because I had assumed what I was viewing was put together by human students or researchers. However, after learning this and reviewing the blog, I think the clean format, welcoming graphics, and interesting recipes outweigh my initial mistrust. Had it been stated outright that the blog was AI-generated, I think I would have been happy to use the blog as a recipe resource without my initial misgivings” (N330).
The most prevalent themes were related to the limitations of AI and/or benefits of humans. Participants’ comments focused on the fact that AI cannot taste and expressed concern that the recipes would not turn out well or would not taste good. Many participants emphasized that they wanted humans to create or test the recipes. Other sentiments indicated that they assumed that the AI-generated recipes would be decent because they were based on existing recipes, or, additionally, that as long as it was a good recipe they did not care if AI created it or not. There was a concern in some comments that, even if AI could create good recipes, AI lacks the ability to optimize recipes like a human could in regard to creativity, diversity, complexity, or refinement. However, other comments indicated that AI may be better at creating recipes because of its ability to scour a large amount of data.
Participants varied in where they drew the line for an acceptable level of AI involvement in a recipe blog. Some indicated that it made no difference if it was AI-generated or not. Some participants were fine if the blog was created by AI, but did not want any AI-generated recipes. Participant L43 wrote, “If I knew that a recipe blog’s photos and layout were AI generated, I would still use the site. However, if the recipe itself is AI generated, then I wouldn’t use the recipe site. It might sound silly, but AI didn’t actually cook and test the recipe- no matter how many fancy words you pour on top of it. And if it’s just copying someone else’s recipe in from a different blog, then why not just use the original?” (L43).
Some comments indicated a strong preference against using AI and highlighted the importance of the human touch in creating recipes and blog posts. Participants wanted to be able to connect to the recipe, the recipe creator, and the “story” that went with it. Participants were concerned that commentary from AI would be disingenuous. Some participants were fine with AI-generated recipes and/or blogs as long as they were verified by humans or had reviews or comments from people who had made the recipes. Some focused only on the quality and did not care about the source. Fewer comments indicated a preference for an AI-generated recipe and/or blog.
Though not as prevalent as other themes, there was a concern for the ethical implications of using AI. Some comments highlighted that it was lazy to use AI. Others voiced the concern that using AI to generate recipes and/or websites would be either stealing content from creators or would prevent users from supporting the content creators themselves.

4. Discussion

The purpose of this cross-sectional study was to assess differences in college students’ ratings of a recipe blog created by dietetic students and a recipe blog that included AI-generated recipes and content. Understanding differences in college student perceptions related to the use of AI-generated discipline-specific knowledge content has implications for curriculum development concerning the use of AI in future dietetics practice.
The participants in this study generally viewed the blogs positively and expressed an intent to use them frequently. There were no other differences in participants’ ratings of the blogs in terms of the ease of recipes, time to prepare the recipes, use of common ingredients, or likelihood of use. The vast majority (approximately 86%) of participants, regardless of which blog they reviewed, reported an intent to utilize the recipe blog at least monthly. This percentage aligns with the Chicory State of Online Recipes report, which indicated that 87% of Gen Z individuals report searching for recipes online [17]). With these positive perceptions, and past evidence that recipe blogs promote healthy behavior changes such as increasing fruit and vegetable consumption [18,19,20], the blogs evaluated in this study have the potential to benefit students and help them overcome barriers to healthy eating including limited time, money, and cooking skills [21,22,23].
The AI-generated blog was rated lower on budget friendliness. This discrepancy was likely because some recipe titles that student researchers selected to be included in the AI-generated blog featured more expensive ingredients such as salmon. However, it is important to note that, while this difference was statistically significant, the small difference likely is not practically significant.
The similar ratings of most criteria correlate with some qualitative comments that indicated some participants could not tell the difference between the student-generated blog and the AI-generated blog. The lack of awareness of the use of AI has been seen in other fields, including mental health. A study by Jain et al. reported that participants rated mental health advice from ChatGPT similar to advice from humans [24].
Nearly one half of participants indicated that knowing a recipe blog was AI-generated would have no impact on how likely they would be to use the blog, supporting the finding that AI-generated content has the possibility to adequately fulfill some users’ needs and meet their expectations. Interestingly, the majority of comments and codes aligned with negative perceptions of AI use. Perhaps when participants indicated their likelihood of using the website if they knew it was AI-generated, it prompted them to reflect more deeply on the possible implications of AI. However, it is important to note that the participants’ perceptions of AI existed on a spectrum. A small percentage of participants (4.4%) indicated being more likely to use a blog if they knew it was AI-generated. Similarly, some participant comments highlighted the potential benefits of AI, including its ability to review vast amounts of information.
Some participants indicated that they did not mind if the blog was AI-generated, but preferred that the use of AI be disclosed. However, Jain et al. reported a decrease in participants’ ratings of the authenticity of AI-generated information when the use of AI was disclosed [24]. It is possible that participants may not be aware that their approval of the blog may decrease if they know that the blog is AI-generated. On the other hand, Jiang et al. describe the positive impact explaining, disclosing, and being transparent regarding the use of and processes followed for AI content generation may have on gaining consumers‘ trust [25].
The main themes from the open-ended responses revealed important concerns about the lack of authenticity, the impersonal tone of the text, as well as the quality and reliability of the AI-generated recipes. Participants emphasized that they preferred recipes that had been tested, which has implications for AI-generated recipes as well as human created recipes. Participants’ comments indicated that recipes with many reviews and comments from individuals who had made the recipes added a perceived credibility to the recipe.
Many participants expressed a preference for evidence of human touch in a recipe and corresponding blog post. Responses included a desire to know that a recipe/blog’s author had real experience and understanding of being a college student, making and eating food, following a recipe, and, while doing so, being a human who makes human mistakes. Many reported that this would increase their ability to trust the content and their motivation to use it. Ho and Chien report that the level of trust consumers have in a blog’s message will affect their consumption behavior [26], highlighting the importance of confidence in trustworthiness of online content, especially when it comes to materials meant to educate and influence diet and other healthy habits. This is consistent with the definition of trust given by Jiang et al., which is that trust is “the willingness of the trustor to engage in an action based on positive expectations regarding the trustee” [25]. Additionally, Ho and Chien explain that the attractiveness of content, which could also be interpreted as the appearance, style, or even the reliability, also aids in a blog’s credibility [26]. This is supported by the findings of the current study that students preferred recipes and blogs that are human in nature, some even commenting that allowing AI to generate such content eliminates the “humanness” that should exist with food. However, some participants were accepting of the blog as long as a human reviewed the content.
Among participants who indicated indifference to using content generated by AI were those who expressed that what really impacted their likelihood of use was whether or not the recipe or blog was of high quality, and that the source (AI or human) was secondary or not important enough to influence use. While a universal definition of high quality was not determined, many participants’ comments utilized words and phrases indicating that what they desired, regardless of who created it, were well-written blogs, easy-to-follow recipes, and good-tasting food as an outcome. It is likely that participants possess varying levels of understanding regarding typical recipe creation and the subsequent quality of food that can be produced when a recipe is followed. Many shared sentiments similar to “a recipe is a recipe”, an idea that the authors interpreted to be the belief that anyone can create a good, functional, or tasty recipe. Other participants, however, expressed doubt about the quality of the recipes/food AI could generate, many speaking to the fact that AI cannot taste. This is supported by previous research that indicates that, due to a lack of human elements or human touch, consumers perceive greater risk when engaging with and using AI-generated content, a factor that negatively impacts their behavioral intent [27].
Participant comments highlighted ethical concerns which align with the creators’ concerns related to copyright infringement and monetary loss in the event that AI-generated content was used instead of work created by human creators [28,29]. The reporting of this concern may have been influenced by the fact that the majority of participants were considered Gen Z, a group of individuals who are known to feel strongly about such social issues [30,31].
The broad recruitment outreach and large sample size are strengths of this study, in addition to the randomized assignment to evaluation groups. The consistent and similar format and design of student and AI-generated blogs also likely eliminated visual or other related bias. Additionally, accounting for confounding factors provided confidence in the identification of the true relationship between the blog recipe type and user ratings.
This study is limited by the low response rate, and failure to collect race and ethnicity data. In addition, participants’ typical AI use and overall perceptions of AI were not assessed. While face validity was determined for the survey, the study was limited because validity and reliability testing was only conducted for the four questions used to assess cooking confidence. This may lower the validity of the study results. Furthermore, the cross-sectional design and only capturing preferences at one point in time weaken the findings. The self-selected nature of the survey may have introduced bias, potentially attracting a higher number of participants with an interest in food or cooking, for example.

5. Conclusions

Though the participants’ recognition of the role of AI-generated content in the blog varied, participants’ comments highlight general recipe blog preferences, such as the appearance and tone of the blog and quality and reliability of the recipes. In a world where the use of AI is becoming prevalent, it is increasingly important to train dietetic students on how to critically engage with and evaluate content created by AI. Students should be aware that, although AI-generated content is readily available to consumers, most consumers can identify AI-generated content and value the personal touch provided by experienced humans. Students should also consider the range of consumer perceptions of AI use in content creation, as well as ethical implications. Students need practice in inserting value-added human perspectives into the digital materials they create to provide quality, evidence-based content for end users.

Author Contributions

Conceptualization, K.N.K., S.L.B., and H.J.W.; methodology, K.N.K., S.L.B., and H.J.W.; formal analysis, K.N.K. and S.M.S.; investigation—K.N.K., S.L.B., S.M.S., M.H.A., B.C.J., C.F., and H.J.W.; writing—original draft preparation, K.N.K., S.L.B., S.M.S., M.H.A., B.C.J., C.F., and H.J.W.; writing—review and editing, K.N.K., S.L.B., S.M.S., M.H.A., B.C.J., C.F., and H.J.W.; supervision, K.N.K., S.L.B., and H.J.W. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

All procedures were reviewed and approved by the Utah State University institutional review board (Protocol #13819); this study was classified as exempt.

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors on request.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Daniel, K.; Msambwa, M.M.; Wen, Z. Can Generative AI Revolutionise Academic Skills Development in Higher Education? A Systematic Literature Review. Eur. J. Educ. 2025, 60, e70036. [Google Scholar] [CrossRef]
  2. Msambwa, M.M.; Wen, Z.; Daniel, K. The impact of AI on the personal and collaborative learning environments in higher education. Eur. J. Educ. 2025, 60, e12909. [Google Scholar] [CrossRef]
  3. US Department of Education, Office of Educational Technology, Artificial Intelligence and Future of Teaching and Learning: Insights and Recommendations, Washington, DC, 2023. Available online: https://files.eric.ed.gov/fulltext/ED631097.pdf (accessed on 27 August 2025).
  4. Chatelan, A.; Clerc, A.; Fonta, P.A. ChatGPT and Future Artificial Intelligence Chatbots: What may be the Influence on Credentialed Nutrition and Dietetics Practitioners? J. Acad. Nutr. Diet. 2023, 123, 1525–1531. [Google Scholar] [CrossRef] [PubMed]
  5. Amiri, M.; Li, J.; Hasan, W. Personalized flexible meal planning for individuals with diet-related health concerns: System design and feasibility validation study. JMIR Form Res. 2023, 7, e46434. [Google Scholar] [CrossRef]
  6. Niszczota, P.; Rybicka, I. The credibility of dietary advice formulated by ChatGPT: Robo-diets for people with food allergies. Nutrition 2023, 112, 112076. [Google Scholar] [CrossRef] [PubMed]
  7. Engrav, C. AI Can Generate Recipes That Can Be Deadly. Food Bloggers Are not Happy. 2024. Available online: https://www.npr.org/2024/09/23/g-s1-23843/artificial-intelligence-recipes-food-cooking-apple (accessed on 26 August 2025).
  8. Sun, Z.; Zhang, Z.; Shen, X.; Zhang, Z.; Liu, Y.; Backes, M.; Zhang, Y.; He, X. Are We in the AI-Generated Text World Already? Quantifying and Monitoring AIGT on Social Media. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Vienna, Austria, 27 July–1 August 2025; pp. 22975–23005. [Google Scholar]
  9. Zhou, J.; Zhang, Y.; Luo, Q.; Parker, A.G.; De Choudhury, M. Synthetic lies: Understanding AI-generated misinformation and evaluating algorithmic and human solutions. In Proceedings of the 2023 CHI Conference on Human Factors in Computing System, Hamburg, Germany, 23–28 April 2023. [Google Scholar] [CrossRef]
  10. Mete, R.; Kellett, J.; Bacon, R.; Shield, A.; Murray, K. The P.O.S.T Guidelines for Nutrition Blogs: A Modified e-Delphi Study. J. Acad. Nutr. Diet. 2021, 121, P1219–P1230. [Google Scholar] [CrossRef]
  11. Nikolaus, C.J.; An, R.; Ellison, B.; Nickols-Richardson, S.M. Food Insecurity among College Students in the United States: A Scoping Review. Adv. Nutr. 2020, 11, 327–348. [Google Scholar] [CrossRef]
  12. Jones, S.A.; Walter, J.; Soliah, L.; Phifer, J.T. Perceived motivators to home food preparation: Focus group findings. J. Acad. Nutr. Diet. 2014, 114, 1552–1556. [Google Scholar] [CrossRef]
  13. Hemmer, A.; Hitchcock, K.; Lim, Y.S.; Butsch Kovacic, M.; Lee, S.Y. Development of Food Literacy Assessment Tool Targeting Adults with Low Income. J. Nutr. Educ. Behav. 2021, 53, 966–976. [Google Scholar] [CrossRef]
  14. Cortina, J.M. What Is coefficient alpha? An examination of theory and applications. J. Appl. Psychol. 1993, 78, 98–104. [Google Scholar] [CrossRef]
  15. Braun, V.; Clarke, V. Using thematic analysis in psychology. Qual. Res. Psychol. 2006, 3, 77–101. [Google Scholar] [CrossRef]
  16. Cresswell, J.W. Qualitative Inquiry and Research Design: Choosing Among Five Approaches, 3rd ed.; Pearson: Boston, MA, USA, 2008. [Google Scholar]
  17. Chicory. The State of Online Recipes. 2024. Available online: https://chicory.co/blog-feed/2024-state-of-online-recipes (accessed on 26 August 2025).
  18. Caplette, M.E.; Provencher, V.; Bissonnette-Maheux, V.; Dugrenier, M.; Lapointe, A.; Gagnon, M.P.; Straus, S.; Desroches, S. Increasing fruit and vegetable consumption through a healthy eating blog: A feasibility study. JMIR Res. Protoc. 2017, 6, e59. [Google Scholar] [CrossRef] [PubMed]
  19. McCully, S.; Don, B.; Updegraff, J. Using the internet to help with diet, weight, and physical activity: Results from the Health Information National Trends Survey (HINTS). J. Med. Internet Res. 2013, 15, e148. [Google Scholar] [CrossRef]
  20. Yen, W.J. Blog-based nutrition education for college students. Food Sci. Technol. 2017, 5, 37–46. [Google Scholar] [CrossRef]
  21. Tyrrell, R.L.; Townshend, T.G.; Adamson, A.J.; Lake, A.A. “I′m not trusted in the kitchen”: Food environments and food behaviours of young people attending school and college. J. Public Health 2016, 38, 289–299. [Google Scholar] [CrossRef]
  22. Hanson, A.J.; Kattelmann, K.K.; McCormack, L.A.; Zhou, W.; Brown, O.N.; Horacek, T.M.; Shelnutt, K.P.; Kidd, T.; Opoku-Acheampong, A.; Franzen-Castle, L.D.; et al. Cooking and meal planning as predictors of fruit and vegetable intake and BMI in first-year college students. Int. J. Environ. Res. Public Health 2019, 16, 2462. [Google Scholar] [CrossRef] [PubMed]
  23. Wongprawmas, R.; Sogari, G.; Menozzi, D.; Mora, C. Strategies to promote healthy eating among university students: A qualitative study using the nominal group technique. Front. Nutr. 2022, 9, 821016. [Google Scholar] [CrossRef]
  24. Jain, G.; Pareek, S.; Carlbring, P. Revealing the source: How awareness alters perceptions of AI and human-generated mental health responses. Internet. Interv. 2024, 36, 100745. [Google Scholar] [CrossRef]
  25. Jiang, X.; Wu, Z.; Yu, F. Constructing consumer trust through artificial intelligence generated content. Acad. J. Bus. Manag. 2024, 6, 263–272. [Google Scholar] [CrossRef]
  26. Ho, H.; Chien, P.H. Influence of message trust in online word-of-mouth on consumer behavior—By the example of food blog. In Proceedings of the 2010 International Conference on Electronics and Information Engineering, Kyoto, Japan, 1–3 August 2010; pp. V1-395–V1-399. [Google Scholar] [CrossRef]
  27. Gonçalves, A.P. Byte by Byte: Influence of Content Generator: Investigating the Impact of AI-Generated Content on Health Outcomes. Master’s Thesis, Universidade NOVA de Lisboa, Lisbon, Portugal, July 2024. [Google Scholar]
  28. Jiang, H.H.; Brown, L.; Cheng, J.; Khan, B.; Gupta, A.; Workman, D.; Hanna, A.; Flowers, J.; Gebru, T. AI art and its impact on artists. In Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society (AIES ‘23); Association for Computing Machinery: New York, NY, USA, 2023; pp. 363–374. [Google Scholar] [CrossRef]
  29. Gaffari, H.; Albarashdi, S. Copyright protection for AI-generated works: Exploring originality and ownership in a digital landscape. Asian J. Int. Law 2025, 15, 23–46. [Google Scholar] [CrossRef]
  30. Seegmiller, S.; Grace, M. Generation Z: Educating and Engaging the Next Generation of Students. About Campus 2017, 23, 21–26. [Google Scholar] [CrossRef]
  31. Kaplan, E.B. The Millennial/Gen Z leftists are emerging: Are sociologists ready for them? Sociol. Perspect. 2020, 63, 408–427. [Google Scholar] [CrossRef]
Dietetics 04 00050 g001
Figure 1. Flowchart of participants.
Figure 1. Flowchart of participants.
Dietetics 04 00050 g001
Table 1. Participant characteristics (N = 948).
Table 1. Participant characteristics (N = 948).
Student-Generated Blog Group (n = 456)AI-Generated Blog Group (n = 492)p-Value
nMean ± SD/%nMean ± SD/%
Age44022.11 ± 3.9047722.03 ± 3.690.764
Female453288 (63.6%)490306 (62.4%)
Male 141 (31.1%) 164 (33.5%)
Non-binary 11 (2.4%) 9 (1.8%)
Other 4 (0.9%) 4 (0.8%)
Prefer not to answer 9 (2%) 7 (1.4%)0.865
Class Rank453 490 0.843
Freshman 90 (19.9%) 97 (19.8%)
Sophomore 79 (17.4%) 85 (17.3%)
Junior 86 (19%) 101 (20.6%)
Senior 107 (23.6%) 121 (24.7%)
Graduate 75 (16.6%) 74 (15.1%)
Other 6 (1.3%) 3 (0.8%)
Don’t know 3 (0.7%) 5 (1%)
Prefer not to answer 7 (1.5%) 4 (0.8%)
Health-related major 41452 (12.6%)45541 (9%)0.10
Taken Food Literacy course 45324 (5.3%)49028 (5.7%)0.887
Cooking Confidence (on a scale from 1 to 5; 5 indicates most confident)
In general, I am confident in my cooking skills.4544.13 ± 4.564914.0 ± 0.930.53
I am confident I can prepare meals using tools I have (eg, stove, oven, knife, pan or pot). *4544.86 ± 6.344914.79 ± 6.100.857
I am confident I can cook foods I have never cooked before if I have a recipe. *4544.44 ± 4.53 4914.81 ± 7.440.362
I am confident I can follow instructions on a food package or recipe. *4544.50 ± 0.70 4904.97 ± 6.070.195
* Items from Food Literacy Assessment Tool (FLitT) [13].
Table 2. Participant assessment of a student-generated blog and an AI-generated blog.
Table 2. Participant assessment of a student-generated blog and an AI-generated blog.
Student-Generated Blog Group (n = 456)AI-Generated Blog Group (n = 492)p-Value
Ratings (on a scale of 1-5) of a Student-Generated Blog and an AI-Generated Blog
Mean ± SDMean ± SD
Budget-Friendly4.14 ± 0.834.02 ± 0.900.025
Easy to Prepare4.13 ±0.844.08 ± 0.780.190
Quick to Prepare3.69 ± 0.983.62 ± 0.940.135
Utilized Common Ingredients3.37 ± 1.063.29 ± 1.070.19
Plan to Cook Recipes at Home3.45 ± 1.163.57 ± 1.200.063
Frequency of Intended Use of the Blog
Never53 (11.6%)61 (12.4%)p-value = 0.574
1–2 time a month244 (53.5%)235 (47.8%)
1–2 times a week119 (26.1%)142 (28.9%)
3–4 times a week40 (6.6%)41 (8.3%)
5–6 times a week2 (0.4%)4 (0.8%)
Every day0 (0%)9 (0.2%)
Prefer not to answer8 (1.8%)8 (1.6%)
Table 3. Spectrum of participants’ perceptions of an AI-generated recipe blog.
Table 3. Spectrum of participants’ perceptions of an AI-generated recipe blog.
ThemeRepresentative Quote
PositiveI love the technology of AI for generating purposes, because I strongly believe that it is a tool that we can use in everyday life with content creation. (N136 1)
Open/Interested/Willing to TryI’ve never tried it before but I would be open (P4)
IndifferentI wouldn’t care if it was AI or not. (N216)
SkepticalIf a recipe was created by AI, I would be skeptical. (L233)
NegativeI dislike the idea of AI so I am biased against it. (N1)
I don’t want to make mustard gas because an AI told me it’d be a cake or something. (L144)
1 Letters and numbers in parentheses indicate participant ID. The first letter indicates participants’ likelihood of using a recipe blog if they knew it was AI-generated; M = more likely, N = no difference, L = less likely.
Table 4. Main themes from free response comments related to participants’ likelihood of using a recipe blog if they knew it was AI-generated.
Table 4. Main themes from free response comments related to participants’ likelihood of using a recipe blog if they knew it was AI-generated.
ThemeRepresentative Quote
Benefits of AII trust AI-generated information more than human generated, as AI-generated can provide less biased data. (M25 1)
I feel as though AI could be more trustworthy in a way because it can take in more information (N334)
Assumed Recipes Would Be Decent Because They Were Based on Existing SourcesRecently I’m realizing more that AI just gathers information and pulls it all together. So i feel like it would generally make decent recipes and I would still try them. (N336)
Previous Experience with AI—Positive I often use AI and find it very reliable. (N213)
I use chatgpt to lookup recipes all the time and it seems to be pretty consistent but faster than information I can gather when trying to find a recipe via pinterest. (N179)
Quality vs. SourceI don’t really care how it’s generated, as long as the user interface is good and the recipes are good. (N172)
If the recipes on an AI-generated blog look tasty, healthy and easy to follow, I’d totally give it a shot. The way the recipes are created doesn’t matter much to me as long as they’re delicious and practical, I’m in! (N8)
Further Verification NeededI think it would be fine, as long as humans could leave reviews. If it’s an AI recipe with four or five stars from humans, I would definitely use it. (N106)
If I knew it were AI-generated I would probably look closer at the measurements for ingredients and be more open to adjusting them to what I think would be better. (N243)
OK if Tested/Verified by HumansI’m fine if AI is involved in the process, but I want the finishing touch to be human. (L101)
AI generated anything would not change my likelihood of using it, of course all things AI generated need to be human checked for quality so we don’t get websites that are unusable. (N69)
AI is “Not There Yet”While i do believe ai can get to the point where it can provide instructions for good food, i don’t think it’s at that point yet. (L129)
AI Is Not Creative/Cannot OptimizeI don’t think AI would be able to come up with new recipes—it could help pull them from the internet, but I wouldn’t trust it to come up with something new. (L246)
An AI-generated recipe is more like a recipe draft; a start, but not a finished, publishable recipe with a good reputation due to repeated trials. (L228)
Benefits of Humans—Prefer Human Creation and/or TestingKnowing that another person has written and actually tested out a recipe before using it makes me more confident that the recipe will turn out how I want it to. (L33)
Benefits of Humans—Prefer Human TouchAI feels less personal. My idea of the website was that it was created to help students who struggle to find and prepare meals for themselves. If the website was generated by AI, I would feel less cared about, in a way. (L91)
From my understanding the reviews were also AI generated, and that makes me feel like I’m being tricked/lied to. (L125)
I think using AI would take away from the humanness of cooking. Kitchens are the metaphorical “heart” of a home. With the use of AI, we loose the deeper understanding of people and their cultures. (L71)
Limitations of AI/Concern for TasteI feel like I wouldn’t trust it as much since the AI has never tasted food. (L42)
AI got no taste buds, I ain’t gonna be skimming through a site that has tasteless recipes or AI-generated ones. (L283)
Limitations of AI—Inaccurate/Unreliable/
Untrustworthy		My personal worry is that an AI-generated blog might not understand how to actually make the food and might end up over cooking something or undercooking something like the meat in a dish. (L127)
Sometimes AI doesn’t get things right. I wouldn’t want it ruining my dinner. (L220)
Previous Experience with AI—NegativeNEVER would I use an AI generated recipe blog. I’ve seen AI make too many weird things. Once I saw an AI generated “vegetarian” recipe that included chicken stock… (L153)
I’ve tried to create AI recipes before with little success so I would be a little wary about the results. (L137)
Ethical ConcernsUsing an AI generated blog seems relatively lazy to me and in my opinion lacks a personal touch. (L55)
I would under no circumstances use a recipe that was AI-generated, because AI technology lifts from recipes written by real people, thereby stealing content and adding competition for ad dollars and other sources of income for these people. (L4)
1 Letters and numbers in parenthesis indicate participant ID. The first letter indicates participants’ likelihood of using a recipe blog if they knew it was AI-generated; M = more likely, N = no difference, L = less likely.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Kraus, K.N.; Bevan, S.L.; Moore Smith, S.; Armstrong, M.H.; Jeppesen, B.C.; Fish, C.; Wengreen, H.J. AI Recipe Blog Is Evaluated Similarly to a Recipe Blog Created by Nutrition and Dietetic Students. Dietetics 2025, 4, 50. https://doi.org/10.3390/dietetics4040050

AMA Style

Kraus KN, Bevan SL, Moore Smith S, Armstrong MH, Jeppesen BC, Fish C, Wengreen HJ. AI Recipe Blog Is Evaluated Similarly to a Recipe Blog Created by Nutrition and Dietetic Students. Dietetics. 2025; 4(4):50. https://doi.org/10.3390/dietetics4040050

Chicago/Turabian Style

Kraus, Katie N., Stacy L. Bevan, Sarah Moore Smith, Maeci H. Armstrong, Brooke Campbell Jeppesen, Catherine Fish, and Heidi J. Wengreen. 2025. "AI Recipe Blog Is Evaluated Similarly to a Recipe Blog Created by Nutrition and Dietetic Students" Dietetics 4, no. 4: 50. https://doi.org/10.3390/dietetics4040050

APA Style

Kraus, K. N., Bevan, S. L., Moore Smith, S., Armstrong, M. H., Jeppesen, B. C., Fish, C., & Wengreen, H. J. (2025). AI Recipe Blog Is Evaluated Similarly to a Recipe Blog Created by Nutrition and Dietetic Students. Dietetics, 4(4), 50. https://doi.org/10.3390/dietetics4040050

Article Metrics

Back to TopTop