Lowering Barriers to Undergraduate Research: Opportunities and Risks of Generative AI at Primarily Undergraduate Institutions

May combine two claims.

Students value mentorship, hands-on practice, and structured support in research culture.

“Dewey et al. (2022) explored students' perceptions of the research culture and documented the value students place on mentorship, hands-on practice, and structured support.”

• cites Dewey et al. (2022) sourcePartly supported

  The Results section, “Students’ Perceived Valuable Learning Experiences” (page unavailable), reports that students valued learning research practices, including planning and running investigations. The Introduction discusses mentorship, and “Implications for Course Design” recommends supports within the curriculum, but the cited passages do not establish that students themselves identified mentorship or structured support as things they valued. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

A GenAI model, given a representative abstract from a civil engineering research article (Rachmad, 2025) with a prompt to summarize key objectives, findings, and implications, produced a concise synthesis that identified the central purpose, major findings, and broader significance.

“The model's output (Figure 2) produced a concise synthesis that identified the central purpose of the study, its major findings, and the broader significance of those findings.”

• cites Rachmad, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Both sets of outputs aligned with current themes in technology, infrastructure, and sustainability, areas commonly emphasized in engineering research.

“Although different in style, both sets of outputs aligned with current themes in technology, infrastructure, and sustainability-areas commonly emphasized in engineering research.”

May combine two claims.

Both ChatGPT and Google Gemini generated concise lists of plausible engineering research topics appropriate for early-stage undergraduate exploration.

“As shown in Figure 1, each tool generated a concise list of plausible engineering research topics appropriate for early-stage undergraduate exploration.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

ChatGPT provided more compact suggestions, while Gemini offered slightly more elaborated descriptions.

“ChatGPT provided more compact suggestions, while Gemini offered slightly more elaborated descriptions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Figure 5 shows a quick guide to understanding regression outputs as explained by ChatGPT.

“Figure 5 shows a quick guide to understanding regression outputs as explained by ChatGPT to demonstrate this scenario.”

Classified as probably not a substantial claim.

ChatGPT is able to highlight grammatical errors, structure, flow, and tone to enhance students' writing.

“Particularly, ChatGPT is able to highlight grammatical errors, structure, flow, and tone to enhance students writing and promote quality writing outputs for each given scenario.”

May combine two claims.

GenAI tools such as ChatGPT and Google Gemini can assist by proposing structured sets of potential topics, identifying emerging trends, and helping students scope projects.

“GenAI tools such as ChatGPT and Google Gemini can assist by proposing structured sets of potential topics, identifying emerging trends, and helping students see what constitutes an appropriately scoped project.”

• AssumptionRests on an assumption: Undergraduate researchers often begin with broad interests but lack experience refining them into feasible research topics.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Both ChatGPT and Google Gemini produced a question and response to collect data from respondents.

“Both ChatGPT and Google Gemini produced a question and response to collect data from the respondents.”

Probably not a substantial claim: may not pass S3 (load-bearing).

ChatGPT and Google Gemini, when instructed to draft three sample survey questions about students and time spent studying compared to academic success, produced three potential survey questions.

“As instructed, the models produced three potential survey questions that could be used for a survey regarding students academic success in conjunction with time spent studying.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

ChatGPT's output was more generic and provided opportunity for further expansion or adaptation, while Google Gemini produced a more sophisticated questionnaire with a short question explanation.

“ChatGPT's output was more generic and provides opportunity for further expansion or adaption of responses, while Google Gemini produced a more sophisticated questionnaire with a short question explanation to explain to the researcher how the question can impact their study.”

Probably not a substantial claim: may not pass S3 (load-bearing).

GenAI tools such as ChatGPT were investigated as supervisory aids for undergraduate and postgraduate researchers.

“et al. (2023) investigated the use of GenAI tools such as ChatGPT as supervisory aids for undergraduate and postgraduate researchers.”

• cites et al. (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Prior work shows growing interest in GenAI as an educational support tool and provides early insights into its benefits and limitations.

“Collectively, prior work shows growing interest in GenAI as an educational support tool and provides early insights into its benefits and limitations.”

Probably not a substantial claim: may not pass S4 (based).

There is limited discussion of how faculty can mentor students in using GenAI responsibly to complement rather than replace research skill development.

“Furthermore, there is limited discussion of how faculty can mentor students in using GenAI responsibly, ensuring that the technology complements rather than replaces the development of research skills.”

May combine two claims.

GenAI may help alleviate structural challenges in undergraduate research at primarily undergraduate institutions, but risks arise when students substitute AI output for authentic engagement.

“These studies highlight structural challenges that GenAI may help alleviate-while also pointing to potential risks when students substitute AI output for authentic engagement with research tasks.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

AI should not replace human supervision but can complement it by providing explanations, examples, and feedback that reinforce faculty guidance.

“Their findings emphasize that AI should not replace human supervision but can complement it by providing explanations, examples, and feedback that reinforce faculty guidance.”

• cites et al. (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Students at primarily undergraduate institutions may encounter barriers including limited prior research experience, a lack of mentorship opportunities, and uncertainty about how to begin independent scholarly work.

“Students at these institutions may also encounter barriers, including limited prior research experience, a lack of mentorship opportunities, and uncertainty about how to begin independent scholarly work (Shamsi et al. 2022).”

• cites Shamsi et al. 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Existing studies do not systematically examine how GenAI can support different stages of the research process or address the unique structural constraints of primarily undergraduate institutions.

“Existing studies do not systematically examine how GenAI can support different stages of the research process, nor do they address the unique structural constraints of PUIs.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

For laboratory or design-based projects, AI can help clarify measurement techniques, data visualization approaches, or error analysis calculations.

“Similarly, for laboratory or design-based projects, AI can help clarify measurement techniques, data visualization approaches, or error analysis calculations.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

By encouraging students to question and refine AI outputs, mentors can use the technology to teach critical evaluation.

“Moreover, by encouraging students to question and refine AI outputs, mentors can use the technology to teach critical evaluation-an essential skill for emerging researchers.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

GenAI tools can provide iterative support throughout the writing process, from brainstorming outlines to refining abstracts, improving transitions, and aligning style with disciplinary conventions.

“GenAI tools can provide iterative support throughout the writing process-from brainstorming outlines to refining abstracts, improving transitions, and aligning style with disciplinary conventions.”

• AssumptionRests on an assumption: Writing is both a key product and learning mechanism in undergraduate research.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

This process also reinforces core research skills such as identifying research gaps, assessing methodological rigor, and connecting findings to broader disciplinary conversations.

“This process also reinforces core research skills such as identifying research gaps, assessing methodological rigor, and connecting findings to broader disciplinary conversations.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

These capabilities allow students to work more independently and gain confidence in interpreting methods sections of published research.

“These capabilities allow students to work more independently and gain confidence in interpreting methods sections of published research.”

Probably not a substantial claim: may not pass S4 (based).

By supporting repetitive or skill-intensive tasks, GenAI tools allow students to focus more deeply on critical thinking, interpretation, and creativity, the intellectual core of research.

“By supporting repetitive or skill-intensive tasks, these tools allow students to focus more deeply on critical thinking, interpretation, and creativity-the intellectual core of research.”

Probably not a substantial claim: may not pass S4 (based).

The literature lacks practical guidance for how students and faculty at primarily undergraduate institutions can integrate GenAI into undergraduate research effectively.

“However, the literature lacks practical guidance for how students and faculty at PUIs can integrate GenAI into undergraduate research effectively.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Used responsibly, GenAI tools encourage revision and self-assessment while improving confidence in scientific communication.

“Used responsibly, these tools encourage revision and self-assessment while improving confidence in scientific communication.”

• AssumptionRests on an assumption: Writing is both a key product and learning mechanism in undergraduate research.

Probably not a substantial claim: may not pass S2 (checkable).

Undergraduate research students can optimize their data collection opportunities through fine tuning of questions, increased data feedback, and initial collection support through the use of GenAI in supplementation with assistance from faculty mentorship.

“Furthermore, undergraduate research students can optimize their data collection opportunities through fine tuning of questions, increased data feedback, and initial collection support through the use of GenAI in supplementation with assistance from faculty mentorship.”

Probably not a substantial claim: may not pass S4 (based).

Faculty mentors can use GenAI to extend their capacity for guidance and supervision despite limited time and resources.

“Likewise, faculty mentors can use GenAI to extend their capacity for guidance and supervision despite limited time and resources.”

Classified as probably not a substantial claim.

Overreliance on GenAI tools may inadvertently limit the development of critical thinking, problem-solving, and technical writing skills in undergraduate research.

“Another concern is the potential for overreliance on GenAI tools, which may inadvertently limit the development of critical thinking, problem-solving, and technical writing skills.”

• AssumptionRests on an assumption: Undergraduate research is intended to promote independent reasoning and intellectual persistence.
• AssumptionRests on an assumption: Students develop strong technical skills during undergraduate research that cannot be replaced with mechanical tools such as GenAI.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

For faculty mentors, AI-generated summaries offer an opportunity to teach students how to verify and evaluate synthesized content.

“For faculty mentors, AI-generated summaries offer an opportunity to teach students how to verify and evaluate synthesized content.”

May combine two claims.

GenAI functions as a collaborative tool that complements, rather than replaces, the mentor-mentee relationship.

“GenAI functions as a collaborative tool that complements, rather than replaces, the mentor-mentee relationship.”

• AssumptionRests on an assumption: Faculty often wish to engage more students in research but face time constraints that limit individualized feedback.

May combine two claims.

Challenges faced by students and faculty at primarily undergraduate institutions often lead to unequal participation in undergraduate research and limit the potential impact of undergraduate research programs at PUIs.

“These challenges often lead to unequal participation in undergraduate research and limit the potential impact of undergraduate research programs at PUIs.”

• AssumptionRests on an assumption: Participation in meaningful research can be difficult to sustain for students and faculty at primarily undergraduate institutions.
• AssumptionRests on an assumption: Faculty at primarily undergraduate institutions often manage heavy teaching loads with limited institutional support, minimal access to graduate assistants, and fewer research-focused resources compared to research-based institutions.

May combine two claims.

GenAI helps students organize ideas, refine arguments, and clarify complex concepts, skills that are often underdeveloped among novice researchers.

“The study highlights the value of GenAI in helping students organize ideas, refine arguments, and clarify complex concepts-skills that are often underdeveloped among novice researchers.”

• cites Walcott (2024)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

GenAI models provide feedback and writing assistance during times when faculty mentors are unavailable to assist undergraduate researchers.

“GenAI models provide the unique ability to provide feedback and writing assistance during times when faculty mentors are unavailable to assist undergraduate researchers.”

• AssumptionRests on an assumption: Faculty often wish to engage more students in research but face time constraints that limit individualized feedback.

Probably not a substantial claim: may not pass S4 (based).

When integrated ethically and strategically, GenAI can help lower the threshold for research engagement, allowing more students to participate in scholarly inquiry earlier and more confidently.

“When integrated ethically and strategically, GenAI can help lower the threshold for research engagement, allowing more students to participate in scholarly inquiry earlier and more confidently.”

May combine two claims.

GenAI tools can help students bridge the gap in literature comprehension by summarizing complex texts, clarifying unfamiliar terminology, and identifying central arguments or recurring themes.

“GenAI tools can help students bridge this gap by summarizing complex texts, clarifying unfamiliar terminology, and identifying central arguments or recurring themes in the literature.”

Probably not a substantial claim: may not pass S4 (based).

Overreliance on GenAI has a detrimental impact on the development of technical writing skills because of the models' quick and functional writing capabilities.

“Technical writing skills can be noted as a having a detrimental impact of development due to overreliance of GenAI due to the quick and functional writing capabilities the models contain.”

• AssumptionRests on an assumption: Students develop strong technical skills during undergraduate research that cannot be replaced with mechanical tools such as GenAI.

Classified as probably not a substantial claim.

The outputs produced by both models can provide a firm baseline of questions that can be further discussed with a faculty mentor or developed further using GenAI and prompt engineering.

“Overall, the outputs produced by both models can provide a firm baseline of questions that can be further discussed with a faculty mentor or developed further using GenAI and prompt engineering for more in-depth responses.”

Classified as probably not a substantial claim. May combine two claims.

GenAI is most beneficial when used in direct alignment with the research process rather than as an isolated tool, and faculty oversight is important for effective learning.

“The authors note that GenAI is most beneficial when used in direct alignment with the research process rather than as an isolated tool, underscoring the importance of faculty oversight to ensure effective learning.”

• cites et al. (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Students who struggle with academic tone or organization can use GenAI as a feedback partner, receiving suggestions on clarity and flow before submitting drafts to mentors.

“Students who struggle with academic tone or organization can use GenAI as a feedback partner, receiving suggestions on clarity and flow before submitting drafts to mentors.”

• AssumptionRests on an assumption: Writing is both a key product and learning mechanism in undergraduate research.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

AI can assist with formatting bibliographies, creating figure captions, and generating presentation summaries, allowing more time for students to focus on content and interpretation.

“AI can also assist with formatting bibliographies, creating figure captions, and generating presentation summaries, allowing more time for students to focus on content and interpretation.”

• AssumptionRests on an assumption: Writing is both a key product and learning mechanism in undergraduate research.

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

The writing feedback opportunity provided by the models promotes quality and ensures time efficiency for student-faculty meetings and learning opportunities.

“The writing feedback opportunity provided by the models not only promotes quality, but also ensures time efficiency for the student -faculty meetings and learning opportunities.”

• AssumptionRests on an assumption: Writing is both a key product and learning mechanism in undergraduate research.

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

GenAI systems can assist with tasks across the research process: brainstorming research questions, identifying and synthesizing literature, explaining research methods, creating visualizations, and improving writing clarity.

“GenAI systems can assist with tasks across the research process: brainstorming research questions, identifying and synthesizing literature, explaining research methods, creating visualizations, and improving writing clarity.”

May combine two claims.

Students conducting a survey-based project can use GenAI to draft sample survey questions, learn about sampling strategies, or receive pseudocode for basic statistical analysis in Python or Excel.

“For instance, students conducting a survey-based project can use GenAI to draft sample survey questions, learn about sampling strategies, or receive pseudocode for basic statistical analysis in Python or Excel.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Comparing a GenAI-generated summary with the original article can prompt productive discussions about accuracy, nuance, and the limits of automated interpretation.

“Comparing the model's summary with the original article can prompt productive discussions about accuracy, nuance, and the limits of automated interpretation.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Structured guidance, iterative feedback, and reflective practice can help mitigate the risk that overreliance on GenAI limits skill development.

“Structured guidance, iterative feedback, and reflective practice can help mitigate this risk.”

Probably not a substantial claim: may not pass S4 (based).

This continuous, low-barrier support helps sustain momentum between mentoring sessions.

“This continuous, lowbarrier support helps sustain momentum between mentoring sessions.”

• AssumptionRests on an assumption: Faculty often wish to engage more students in research but face time constraints that limit individualized feedback.

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Such summaries can help students understand the overall direction of the research and build confidence before engaging in deeper analysis.

“While not a substitute for reading the full article, such summaries can help students understand the overall direction of the research and build confidence before engaging in deeper analysis.”

Probably not a substantial claim: may not pass S4 (based).

By assisting with idea generation, background research, or writing preparation, GenAI can help first- and second-year students participate meaningfully in research projects earlier in their academic careers.

“By assisting with idea generation, background research, or writing preparation, these tools can help first-and second-year students participate meaningfully in research projects earlier in their academic careers.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Emphasizing digital literacy equips students with critical evaluation skills necessary for both academic and professional contexts.

“Emphasizing digital literacy also equips students with critical evaluation skills necessary for both academic and professional contexts.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

When used responsibly, GenAI technologies can foster student confidence, streamline faculty mentorship, and expand access to high-impact learning experiences in undergraduate research.

“When used responsibly, these technologies can foster student confidence, streamline faculty mentorship, and expand access to high-impact learning experiences that define the undergraduate research model.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Training workshops for both faculty and students can foster shared understanding of ethical standards, citation practices, and data privacy considerations.

“Training workshops for both faculty and students can foster shared understanding of ethical standards, citation practices, and data privacy considerations.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Providing shared access to vetted AI platforms and incorporating digital literacy instruction into research training will help democratize participation in undergraduate research while maintaining consistent ethical and educational standards.

“Doing so will help democratize participation in undergraduate research while maintaining consistent ethical and educational standards across diverse learning environments.”

• AssumptionRests on an assumption: Disparities persist in access to premium AI features, reliable internet, and institutional licensing.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Integrating GenAI training into first-year research or methods courses can institutionalize equitable access while reinforcing responsible technology use as part of the broader educational mission.

“Integrating GenAI training into first-year research or methods courses can institutionalize equitable access while reinforcing responsible technology use as part of the broader educational mission.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Students may unintentionally submit AI-generated material as original work or share sensitive research data with public AI systems that store or reuse inputs for model training.

“Students may unintentionally submit AI-generated material as original work or share sensitive research data with public AI systems that store or reuse inputs for model training.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Providing campus-wide access to approved AI tools, along with digital ethics education, will help level the playing field for students across disciplines.

“Providing campus-wide access to approved AI tools, along with digital ethics education, will help level the playing field for students across disciplines.”

• AssumptionRests on an assumption: Disparities persist in access to premium AI features, reliable internet, and institutional licensing.

Probably not a substantial claim: may not pass S4 (based).

GenAI can make research experiences more attainable for students who might otherwise lack the time, confidence, or background knowledge to engage.

“GenAI can make research experiences more attainable for students who might otherwise lack the time, confidence, or background knowledge to engage.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Students at smaller or under-resourced institutions may have limited exposure to advanced AI tools, exacerbating existing inequities in research opportunity.

“Students at smaller or under-resourced institutions may have limited exposure to advanced AI tools, exacerbating existing inequities in research opportunity.”

• AssumptionRests on an assumption: Disparities persist in access to premium AI features, reliable internet, and institutional licensing.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

AI-generated reading summaries, example datasets, or feedback templates can supplement one-on-one guidance.

“For instance, AI-generated reading summaries, example datasets, or feedback templates can supplement one-on-one guidance,”

Probably not a substantial claim: may not pass S4 (based).

At institutions where faculty balance substantial teaching loads, GenAI can extend mentorship capacity by automating certain instructional supports.

“At institutions where faculty balance substantial teaching loads, GenAI can extend mentorship capacity by automating certain instructional supports.”

• AssumptionRests on an assumption: Primarily undergraduate institutions often operate with limited resources, smaller faculty teams, and heavy teaching responsibilities.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Thoughtful adoption of GenAI can help bridge gaps at primarily undergraduate institutions-expanding access to research opportunities, supporting faculty mentorship, and promoting equitable participation-provided that institutions implement structured frameworks for ethical and educational use.

“Thoughtful adoption of GenAI can help bridge these gaps-expanding access to research opportunities, supporting faculty mentorship, and promoting equitable participationprovided that institutions implement structured frameworks for ethical and educational use.”

• AssumptionRests on an assumption: Common barriers at primarily undergraduate institutions include limited prior experience, lack of methodological familiarity, time constraints, and limited faculty availability.
• AssumptionRests on an assumption: Primarily undergraduate institutions often operate with limited resources, smaller faculty teams, and heavy teaching responsibilities.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

By normalizing transparency about AI assistance, institutions can prevent misuse while promoting responsible innovation in academic practice.

“By normalizing transparency about AI assistance, institutions can prevent misuse while promoting responsible innovation in academic practice.”