Stanford faculty weigh in on ChatGPT’s shake-up in education Stanford Graduate School of Education

As of July 2023, it is free to those who sign up for an account using an email address, Google, Microsoft, or Apple account.

The way AI technology is booming in every sphere of life, the day when quality education will be more easily accessible is not far. For example, Georgia Tech has created an adaptive learning platform for its computer science master’s program. This platform uses AI to personalize the learning experience for each student. Similarly, Stanford has its own AI Laboratory, where researchers work on cutting-edge AI projects. MIT is also heavily invested in AI with its MIT Intelligence Quest (MIT IQ) and MIT-IBM Watson AI Lab initiatives, exploring the potential of AI in various fields. Relations graph of pedagogical roles and objectives for implementing chatbots.

The design of CPAs must consider social, emotional, cognitive, and pedagogical aspects (Gulz et al., 2011; King, 2002). Nonetheless, the existing review studies have not concentrated on the chatbot interaction type and style, the principles used to design the chatbots, and the evidence for using chatbots in an educational setting. The traditional education system faces several issues, including overcrowded classrooms, a lack of personalized attention for students, varying learning paces and styles, and the struggle to keep up with the fast-paced evolution of technology and information.

However, there have been contradictory findings related to critical thinking, learning engagement, and motivation. Deng and Yu (2023) found that chatbots had a significant and positive influence on numerous learning-related aspects but they do not significantly improve motivation among students. Contrary, Okonkwo and Ade-Ibijola (Okonkwo & Ade-Ibijola, 2021), as well as (Wollny et al., 2021) find that using chatbots increases students’ motivation. To this end, initial codes were identified by open coding and iteratively improved through comparison, group discussion among the authors, and subsequent code expansion. Further, codes were supplemented with detailed descriptions until a saturation point was reached, where all included studies could be successfully mapped to codes, suggesting no need for further refinement. As an example, codes for RQ2 (Pedagogical Roles) were adapted and refined in terms of their level of abstraction from an initial set of only two codes, 1) a code for chatbots in the learning role and 2) a code for chatbots in a service-oriented role.

In general, the followed approach with these chatbots is asking the students questions to teach students certain content. Moreover, it has been found that teaching agents use various techniques to engage students. Chatbots have been found to play various roles in educational contexts, which can be divided into four roles (teaching agents, peer agents, teachable agents, and peer agents), with varying degrees of success (Table 6, Fig. 6).

The Peril and Promise of Chatbots in Education

In view of that, it is worth noting that the embodiment of ECs as a learning assistant does create openness in interaction and interpersonal relationships among peers, especially if the task were designed to facilitate these interactions. Despite these insights, there remains a significant gap in the literature regarding a comprehensive understanding of teachers’ and students’ perceptions of AICs, particularly in how these perceptions influence their acceptance and effectiveness in language education. This gap is more pronounced in understanding how the design and linguistic features of AICs impact user satisfaction and engagement. While studies like those of Chen et al. (2020) and Chocarro et al. (2023) have begun exploring these areas, there is a need for a more targeted framework to evaluate satisfaction with AICs in the context of language learning.

• Concerning the design principles behind the chatbots, slightly less than a third of the chatbots used personalized learning, which tailored the educational content based on learning weaknesses, style, and needs.
• Chatbots may be better at tutoring certain subjects than others, so be sure to try it out first to assess the helpfulness of the responses.
• “Second, teachers can use the tool as a way of generating many examples and nonexamples of a form or genre.
• For instance, Winkler and Söllner (2018) classified the chatbots as flow or AI-based, while Cunningham-Nelson et al. (2019) categorized the chatbots as machine-learning-based or dataset-based.
• You might first use the chatbot to help you define a project and break down the work into manageable chunks, then clarify the function or routine you want to work on.
• Three categories of research gaps were identified from empirical findings (i) learning outcomes, (ii) design issues, and (iii) assessment and testing issues.

They offer students guidance, motivation, and emotional support—elements that AI cannot completely replicate. Bii (2013) defined educational chatbots as chatbots conceived for explicit learning objectives, whereas Riel (2020) defined it as a program that aids in achieving educational and pedagogical goals but within the parameters of a traditional chatbot. Empirical studies have positioned ECs as a personalized teaching assistant or learning partner (Chen et al., 2020; Garcia Brustenga et al., 2018) that provides scaffolding (Tutor Support) through practice activities (Garcia Brustenga et al., 2018). They also support personalized learning, multimodal content (Schmulian & Coetzee, 2019), and instant interaction without time limits (Chocarro et al., 2021).

SYSTEMATIC REVIEW article

Considering Microsoft’s extensive integration efforts of ChatGPT into its products (Rudolph et al., 2023; Warren, 2023), it is likely that ChatGPT will become widespread soon. Educational institutions may need to rapidly adapt their policies and practices to guide and support students in using educational chatbots safely and constructively manner (Baidoo-Anu & Owusu Ansah, 2023). Educators and researchers must continue to explore the potential benefits and limitations of this technology to fully realize its potential.

• None of the AICs reached the desired level of conversational naturalness, as participants found their responses predictable and lacking the adaptability seen in human tutors.
• The SD values show a similar level of variation in the weekly interaction hours across all four AICs for both Spanish and Czech participants, suggesting a comparable spread of interaction frequencies within each group.
• Interestingly, 38.46% (5) of the journal articles were published recently in 2020.
• They should ensure that the information they provide and how they use the model aligns with ethical standards and legal obligations.

Chatbots can help educational institutions in data collection and analysis in various ways. Firstly, they can collect and analyze data to offer rich insights into student behavior and performance to help them create more effective learning programs. Secondly, chatbots can gather data on student interactions, feedback, and performance, which can be used to identify areas for improvement and optimize learning outcomes. Thirdly education chatbots can access examination data and student responses in order to perform automated assessments. The bots can then process this information on the instructor’s request to generate student-specific scorecards and provide learning gap insights.

Bard, a generative AI chatbot developed by Google, relies on the Pathways Language Model (PaLM) large language model. Remember to read the terms of service of the tool when deciding to access it. Some chatbots have options to opt out of sharing data which are described in the terms of service. LL provided a concise overview of the existing literature and formulated the methodology. All three authors collaborated on the selection of the final paper collection and contributed to crafting the conclusion.

They will play an increasingly vital role in personalized learning, adapting to individual student preferences and learning styles. Moreover, chatbots will foster seamless communication between educators, students, and parents, promoting better engagement and learning outcomes. AI chatbots equipped with sentiment analysis capabilities can play a pivotal role in assisting teachers. By comprehending student sentiments, these chatbots help educators modify and enhance their teaching practices, creating better learning experiences. Promptly addressing students’ doubts and concerns, chatbots enable teachers to provide immediate clarifications, fostering a more conducive and effective learning environment.

Simultaneously, rendering the AICs’ voice generation more human-like can be attained through more sophisticated Text-to-Speech (TTS) systems that mimic the intonation, rhythm, and stress of natural speech (Jeon et al., 2023). The second dimension of the CHISM model, focusing on the Design Experience (DEX), underscores its critical role in fostering user engagement and satisfaction beyond the linguistic dimension. Elements such as the chatbot interface and multimedia content hold substantial importance in this regard. An intuitive and user-friendly interface enriches the overall user experience and encourages interaction (Chocarro et al., 2021; Yang, 2022). Additionally, the incorporation of engaging multimedia content, including videos, images, and other emerging technologies, can also increase users’ attention and engagement (Jang et al., 2021; Kim et al., 2019). AI-powered chatbots can help automate assessment processes by accessing examination data and learner responses.

Interestingly, no feedback from the EC group mentioned difficulties in using the EC nor complexity in interacting with it. It was presumed that students welcomed such interaction as it provided learning support and understood its significance. As for the qualitative findings, firstly, even though the perception of learning did not show much variation statistically, the EC group showed additional weightage that implicates group activities, online feedback, and interaction with the lecturer as impactful. Interestingly, the percentage of students that found “interaction with lecturer” and “online feedback and guidance” for the EC was higher than the control group, and this may be reflected as a tendency to perceive the chatbot as an embodiment of the lecturer. Furthermore, as for constructive feedback, the outcomes for both groups were very similar as the critiques were mainly from the teammates and the instructor, and the ECs were not designed to critique the project task.

The teaching agent simply mimics a tutor by presenting scenarios to be discussed with students. In other studies, the teaching agent emulates a teacher conducting a formative assessment by evaluating students’ knowledge with multiple-choice questions (Rodrigo et al., 2012; Griol et al., 2014; Mellado-Silva et al., 2020; Wambsganss et al., 2020). The implications of the research findings for policymakers and researchers are extensive, shaping the future integration of chatbots in education.

And if it’s asked about something outside of its areas of expertise, it will tell users it can’t help with the question, instead of making something up, a characteristic of most chatbots that pull information from the entire internet. Instead of absorbing information from the entire internet to train its artificially intelligent brain, Stretch is only learning on materials that have been developed or vetted by ISTE and ASCD. Eventually, the tool may include information from other education and tech organizations that ISTE partners with. It was observed that communicating merely was not the main priority anymore as cooperation towards problem-solving is of utmost importance. Example feedback is such as “I learn to push myself more and commit to the project’s success.” Nevertheless, in both groups, all the trends are almost similar.

Their favorite music is being streamed from distant servers, directly to their smart device. Unfortunately, in many public schools in the United States and internationally, printed textbooks, and lecturing to large groups of students are the only available teaching methods. That will enable Stretch to avoid the pitfalls of ChatGPT and similar chatbots, which often spit out inaccurate or outdated information, said Richard Culatta, ISTE’s CEO, during a roundtable discussion and demonstration with reporters here. (For instance, a chatbot mimicking President Barack Obama inaccurately parroted his administration’s critics as his own views when talking to a reporter about the president’s record on K-12 education).

None of the studies discussed the platforms on which the chatbots run, while only one study (Wollny et al., 2021) analyzed the educational roles the chatbots are playing. The study used “teaching,” “assisting,” and “mentoring” as categories for educational roles. This study, however, uses different classifications (e.g., “teaching agent”, “peer agent”, “motivational agent”) supported by the literature in Chhibber and Law (2019), Baylor (2011), and Kerlyl et al. (2006). Other studies such as (Okonkwo and Ade-Ibijola, 2021; Pérez et al., 2020) partially covered this dimension by mentioning that chatbots can be teaching or service-oriented. I believe the most powerful learning moments happen beyond the walls of the classroom and outside of the time boxes of our course schedules. Authentic learning happens when a person is trying to do or figure out something that they care about — much more so than the problem sets or design challenges that we give them as part of their coursework.

The study by Pérez et al. (2020) reviewed the existing types of educational chatbots and the learning results expected from them. Smutny and Schreiberova (2020) examined chatbots as a learning aid for Facebook Messenger. Thomas (2020) discussed the benefits of educational chatbots for learners and educators, showing that the chatbots are successful educational tools, and their benefits outweigh the shortcomings and offer a more effective educational experience.

Only four chatbots (11.11%) used a user-driven style where the user was in control of the conversation. A user-driven interaction was mainly utilized for chatbots teaching a foreign language. Shows that ten (27.77%) articles presented general-purpose educational chatbots that were used in various educational contexts such as online courses (Song et al., 2017; Benedetto & Cremonesi, 2019; Tegos et al., 2020). The approach authors use often relies on a general knowledge base not tied to a specific field. For instance, Winkler and Söllner (2018) classified the chatbots as flow or AI-based, while Cunningham-Nelson et al. (2019) categorized the chatbots as machine-learning-based or dataset-based.

Moreover, according to Cunningham-Nelson et al. (2019), one of the key benefits of EC is that it can support a large number of users simultaneously, which is undeniably an added advantage as it reduces instructors’ workload. Colace et al. (2018) describe ECs as instrumental when dealing with multiple students, especially testing behavior, keeping track of progress, and assigning tasks. Furthermore, ECs were also found to increase autonomous learning skills and tend to reduce the need for face-to-face interaction between instructors and students (Kumar & Silva, 2020; Yin et al., 2021). Conversely, this is an added advantage for online learning during the onset of the pandemic. Likewise, ECs can also be used purely for administrative purposes, such as delivering notices, reminders, notifications, and data management support (Chocarro et al., 2021).

The Assisting role is the support in terms of simplifying learners’ everyday life, e.g. by providing opening times of the library. The Mentoring role is the support in terms of students’ personal development, e.g. by supporting Self-Regulated Learning. From a pedagogical standpoint, all three roles are essential for learners and should therefore be incorporated in chatbots. These pedagogical roles are well aligned with the four implementation objectives reported in RQ1. While Skill Improvement and Students’ Motivation is strongly related to Learning, Efficiency of Education is strongly related to Assisting. The Mentoring role instead, is evenly related to all of the identified objectives for implementing chatbots.

Chatbots’ responses can vary in accuracy, and there is a risk of conveying incorrect or biased information. Universities must ensure quality control mechanisms to verify the accuracy and reliability of the AI-generated content. Special care must be taken in situations where faulty information could be dangerous, such as in chemistry laboratory experiments, using tools, or constructing mechanical devices or structures. Here, we discuss some of the advantages, opportunities, and challenges of chatbots in primary, secondary, and higher education. You can foun additiona information about ai customer service and artificial intelligence and NLP. It should be noted that sometimes chatbots fabricate information, a process called “hallucination,” so, at least for the time being, references and citations should be carefully verified.

Most schools and universities have upgraded their feedback collection process by shifting from print to online forms. While chatting with bots, students will have the chance to explain their claims. On the other hand, the bot can be trained to ask additional questions based on their previous answers. The research, conducted over two academic years (2020–2022) with a mixed-methods approach and convenience sampling, initially involved 163 students from the University of X (Spain) and 86 from the University of X (Czech Republic).

Data extraction strategy

Much like a dedicated support system, they tirelessly cater to the needs of both students and teachers, providing prompt responses and assistance at any time, day or night. This kind of availability ensures that learners and educators can access essential information and support whenever they need it, fostering a seamless and uninterrupted learning experience. To delineate and map the field of chatbots in education, initial findings were collected by a preliminary literature search.

However, the final participant count was 155 Spanish students and 82 Czech students, as some declined to participate or did not submit the required tasks. Participation was voluntary, and students who actively engaged with the chatbots and completed all tasks, including submitting transcripts and multiple-date screenshots, were rewarded with extra credits in their monthly quizzes. This approach ensured higher participation and meaningful interaction with the chatbots, contributing to the study’s insights into the effectiveness of AICs in language education. A chatbot, short for chatterbot, is a computer program that uses artificial intelligence (AI) to conduct a conversation via auditory or textual methods and interacts with humans in their natural languages. These interactions usually occur through websites, messaging applications, or mobile apps, where the bot is capable of simulating and maintaining human-like conversations and perform different tasks (Adamopoulou & Moussiades, 2020).

AI-powered chatbots are designed to mimic human conversation using text or voice interaction, providing information in a conversational manner. Chatbots’ history dates back to the 1960s and over the decades chatbots have evolved significantly, driven by advancements in technology and the growing demand for automated communication systems. Created by Joseph Weizenbaum at MIT in 1966, ELIZA was one of the earliest chatbot programs (Weizenbaum, 1966). ELIZA could mimic human-like responses by reflecting user inputs as questions. Another early example of a chatbot was PARRY, implemented in 1972 by psychiatrist Kenneth Colby at Stanford University (Colby, 1981). PARRY was a chatbot designed to simulate a paranoid patient with schizophrenia.

Stretch is one of the first so-called “walled garden AI” tools trained on a limited, carefully curated pool of information to serve a specific community, in education or any area, Culatta said. ISTE is still developing Stretch and hopes to give a wider group of educators a chance to use the tool soon. Eventually, Stretch may be used to help educators with research and professional development.

For instance, Okonkwo and Ade-Ibijola (2021) found out that chatbots motivate students, keep them engaged, and grant them immediate assistance, particularly online. Additionally, Wollny et al. (2021) argued that Chat PG educational chatbots make education more available and easily accessible. To summarize, incorporating AI chatbots in education brings personalized learning for students and time efficiency for educators.

Studies that used questionnaires as a form of evaluation assessed subjective satisfaction, perceived usefulness, and perceived usability, apart from one study that assessed perceived learning (Table 11). Assessing students’ perception of learning and usability is expected as questionnaires ultimately assess participants’ subjective opinions, and thus, they don’t objectively measure metrics such as students’ learning. In general, the studies conducting evaluation studies involved asking participants to take a test after being involved in an activity with the chatbot.

Therefore, (Goal 4) of our review lies in the investigation of the adaptation approaches used by chatbots in education. For (Goal 5), we want to extend the work of (Winkler and Soellner, 2018) and (Pérez et al., 2020) regarding Application Clusters (AC) and map applications by further investigating specific learning domains in which chatbots have been studied. Addressing these gaps in the existing literature would significantly benefit the field of education. Firstly, further research on the impacts of integrating chatbots can shed light on their long-term sustainability and how their advantages persist over time.

Moreover, chatbots may interact with students individually (Hobert & Meyer von Wolff, 2019) or support collaborative learning activities (Chaudhuri et al., 2009; Tegos et al., 2014; Kumar & Rose, 2010; Stahl, 2006; Walker et al., 2011). Chatbot interaction is achieved by applying text, speech, graphics, haptics, gestures, and other modes of communication to assist learners in performing educational tasks. From the viewpoint of educators, integrating AI chatbots in education brings significant advantages. AI chatbots provide time-saving assistance by handling routine administrative tasks such as scheduling, grading, and providing information to students, allowing educators to focus more on instructional planning and student engagement. Educators can improve their pedagogy by leveraging AI chatbots to augment their instruction and offer personalized support to students. By customizing educational content and generating prompts for open-ended questions aligned with specific learning objectives, teachers can cater to individual student needs and enhance the learning experience.

Another example is the E-Java Chatbot (Daud et al., 2020), a virtual tutor that teaches the Java programming language. While the identified limitations are relevant, this study identifies limitations from other perspectives such as the design of the chatbots and the student experience with the educational chatbots. To sum up, Table 2 shows some gaps that this study aims at bridging to reflect on educational chatbots in the literature. Several studies have found that educational chatbots improve students’ learning experience.

Remember to take the lead when using chatbots for team projects, making your own choices while incorporating the helpful and discarding what is not. Metacognitive skills can help students understand how learning works, increase awareness of gaps in their learning, and lead them to develop study techniques (Santascoy, 2021). Stanford has academic skills coaches that support students in developing metacognitive and other skills, but you might also integrate metacognitive activities into your courses with the assistance of an AI chatbot. For example, you and your students could use a chatbot to reflect on their experience working on a group project or to reflect on how to improve study habits.

For instance, both groups portrayed high self-realization of their value as a team member at the end of the course, and it was deduced that their motivational belief was influenced by higher self-efficacy and intrinsic value. Next, in both groups, creativity was overshadowed by post-intervention teamwork significance. Therefore, we conclude that ECs significantly impact learning performance and teamwork, but affective-motivational improvement may be overshadowed by the homogenous learning process for both groups. Firstly, Kearney et al. (2009) explained that in homogenous teams (as investigated in this study), the need for cognition might have a limited amount of influence as both groups are required to be innovative simultaneously in providing project solutions. Lapina (2020) added that problem-based learning and solving complex problems could improve the need for cognition.

Adopting EUD tools to build chatbots would accelerate the adoption of the technology in various fields. In terms of the educational role, slightly more than half of the studies used teaching agents, while 13 studies (36.11%) used peer agents. Only two studies presented a teachable agent, and another two studies presented a motivational agent. Teaching agents gave students tutorials or asked them to watch videos with follow-up discussions.

Based on my initial explorations of the current capabilities and limitations of both types of chatbots, I opted for scripted chatbots. Most learning happens in the 99.9% of our lives when we are not in a classroom. The COVID-19 pandemic pushed educators and students out of their classrooms en masse.

This learning concept involves repeating the old lessons, just before you forget them. The spaced interval learning was used as a basis for developing an app that helps people to track the learning process and reminds them to repeat the lessons they are about to forget. The app was created by the Polish inventor Piotr Wozniak and promoted by the SuperMemo company.

This method encourages students to ask questions and actively participate in processes comfortably. As a result, it significantly increases concentration level and comprehensive understanding. The success of chatbot implementation depends on how easily educatee perceive https://chat.openai.com/ and adapt to their use. If they find tools complex or difficult to navigate, it may hinder their acceptance and application in educational settings. Ensuring a user-friendly interface and straightforward interactions is important for everyone’s convenience.

One important limitation to be mentioned here is the exclusion of alternative keywords for our search queries, as we exclusively used chatbot as keyword in order to avoid search results that do not fit our research questions. A second limitation may lie in the formation of categories and coding processes applied, which, due to the novelty of the findings, could not be built upon theoretical frameworks or already existing code books. Although we have focused on ensuring that codes used contribute to a strong understanding, the determination of the abstraction chatbot in education level might have affected the level of detail of the resulting data representation. While Mentoring chatbots to support Self-Regulated Learning are intended to encourage students to reflect on and plan their learning progress, Mentoring chatbots to support Life Skills address general student’s abilities such as self-confidence or managing emotions. Finally, Mentoring chatbots to support Learning Skills, in contrast to Self-Regulated Learning, address only particular aspects of the learning process, such as new learning strategies or helpful learning partners.

In this type of support, the student himself is the focus of the conversation and should be encouraged to plan, reflect or assess his progress on a meta-cognitive level. One example is the chatbot in (Cabales, 2019), which helps students develop lifelong learning skills by prompting in-action reflections. None of the articles explicitly relied on usability heuristics and guidelines in designing the chatbots, though some authors stressed a few usability principles such as consistency and subjective satisfaction. Further, none of the articles discussed or assessed a distinct personality of the chatbots though research shows that chatbot personality affects users’ subjective satisfaction. Concerning the design principles behind the chatbots, slightly less than a third of the chatbots used personalized learning, which tailored the educational content based on learning weaknesses, style, and needs. Other chatbots used experiential learning (13.88%), social dialog (11.11%), collaborative learning (11.11%), affective learning (5.55%), learning by teaching (5.55%), and scaffolding (2.77%).

The study found similar results in both settings, strengthening the argument for the broader relevance and potential of AICs in diverse educational environments. The landscape of mobile-application language learning (MALL) has been significantly reshaped in recent years with the incorporation of AICs (Pham et al., 2018). This innovative approach to mobile learning has been positively received by both students and teachers. For example, Chen et al. (2020) highlighted the effectiveness of AICs for Chinese vocabulary learning by comparing chatbot-based tutoring with traditional classroom settings. The study reported positive user feedback on the chatbot’s ease of use, usefulness, and enjoyment, as measured by the Technology Acceptance Model (TAM). Similarly, Yang (2022) underscored the favourable views of AICs in English language education, with teachers valuing the chatbot’s capacity to manage routine tasks, thereby allowing them to concentrate on more substantial classroom duties.…