Conversational AI, particularly text-generating large language models (LLMs) are all the rage right now. The poster child of this type of generative AI, ChatGPT, has since its launch become the fastest adopted consumer technology application ever (Chow 2023), surpassing the initial growth rates of e.g. Instagram and TikTok.

Similar tools have emerged alongside ChatGPT: Quora’s Poe, Bing’s generative search interface, Rytr, Kactus.ai, Character.ai.. The list goes on. To make the most of natural language based user interfaces (UIs) like ChatGPT, companies are hiring prompt engineers – a job title that did not exist a year ago – and integrating generative AI capabilities into existing product suites. For example, Salesforce, one of the biggest customer relationship management (CRM) technology stacks in the world, recently announced its integration with ChatGPT to turbo-charge its existing AI, Einstein.

In the context of higher education, the popularisation of text-generating large language models has not gone unnoticed. Amid fears of an avalanche of computer-generated “cheating”, universities across the world have scrambled together guidelines for when and how generative AI applications like ChatGPT are allowed to be used in and outside of the classroom.

Most recently, the University of Helsinki (2023) published its guidelines for using AI in teaching and learning. The key takeaway seems to be that generative AI applications can be used in assignments and other reports, but that its usage should be disclosed, including the AI model or consumer interface used along with date and prompts used for generating the text. Students are also encouraged to critically reflect on the usefulness and potential biases of AI.

Many sides of conversational generative AI

Text-generating novel AI tools have the potential to disrupt the current status quo in higher education in many ways, but it also comes with significant benefits and challenges (Dwivedi et al. 2023).

In terms of benefits, generative AI can be used to for example create customised learning experiences that cater to the needs and preferences of individual learners. This can help students to learn more efficiently and effectively. Generative AI can also be used to help with proofreading and other language-editing tasks. It can be of help in making text more concise or improving readability, as well as in generating new ideas and creating original content, which can enhance students’ creativity and critical thinking skills.

Finally, generative AI can be used to create more accurate and reliable assessments of students’ learning outcomes, enabling teachers to better understand students’ strengths and weaknesses.

On the other hand, the use of generative AI in education raises ethical concerns around issues such as privacy, data security, and bias. There is a risk that students’ personal information could be compromised, or that AI algorithms could perpetuate existing biases and discrimination due to lack of transparency as to how and why a given output is generated.

Further, there is a risk that generative AI could generate low-quality or inaccurate content, which could negatively impact students’ learning outcomes. Most of current iterations of consumer-facing conversational generative AI applications cannot accurately list the sources used for generating text. Finally, utilising generative AI tools doesn’t necessarily teach students critical thinking skills if too much of problem-solving is outsourced to AI.

In Haaga-Helia’s bachelor’s course “Creativity and Innovation in Hospitality”, we sought to test how generative AI could be integrated into the classroom.

Conversational generative AI in the classroom

During the implementation planning an idea was born about utilizing and testing AI in an educational setting. Students were given an assignment to be completed in pairs where the main requirement was to familiarise themselves with and critically analyse some of the biggest innovations in tourism and hospitality.

They were first asked to read an article about 100 Innovations That Transformed Tourism (Hjalager, 2015). Based on the reading, they were asked to write a blog post where they had to identify and critically assess five innovations that they believe had the biggest impact on tourism and hospitality and drawing on the broader expertise on the topic, propose two new innovations that have emerged in recent years.

After the assignments had been submitted, extended activity in class was held, where we utilised different generative conversational AI tools. In the first part of the activity, students were asked to give the AI a prompt of a persona of their choice, and as that persona, ask them to propose two new innovations. In our case we gave them 3 options they had to utilize for generating suggestions for new innovations: 1) choosing older innovator’s persona; 2) choosing new age innovator’s persona; and 3) picking persona of their own choice.

After they had gotten all six answers, they compared them to their own suggestions and reflected on the differences and real-world applicability of the proposed innovations in the hospitality and tourism industry. Finally, we ended the activity by asking students to feed the AI tool of their choice their initial blog post (before the AI activity) and ask it to improve their writing as well as reflect on how they used the tool and what prompts were used in the process. Final part of the activity was reflection on the pros and cons of conversational generative AI.

In our assessment, we concentrated on how well students were able to follow the structure as well as new proposals and critical assessment and justification of new innovations that emerged as a result of the activity. We were particularly looking for students to evidence an ability to build a convincing argument as well as to showcase their creative usage of AI generator(s).

After the activity, we collected feedback via a survey (n=21). The results shed interesting light on the current use of conversational generative AI in the classroom. First, we asked students to rate their familiarity with conversational AI like ChatGPT. Against our assumptions, 67 % said they hadn’t used conversational generative AI before, and only one student said they had used applications like ChatGPT already “a lot”.

Use-cases included asking conversational AI for topic suggestions for assignments or other projects, using AI to paraphrase and re-phrase text, using AI to correct grammatic errors and to sound more professional, as well as using AI to help come up with out-of-the-box ideas. When asked about the usefulness of conversational generative AI, the majority of students (81 %) said that it is useful or very useful.

Similarly, when asking students to rate on a scale of 1-5 the ease-of-use of conversational generative AI, the mean was 3.8 and median answer 4. Finally, when asked whether students thought our learning activity was fun or not, the majority (86 %) said they enjoyed it, with a rating of 4.3 (out of 5); median answer 5.

To embrace or to shy away from conversational generative AI

Conversational generative AI has the potential to transform higher education in many positive ways, but it is important to carefully consider the benefits and challenges of using the technology before implementing it in university settings. In our case, we had the clear vision of how to utilize the chat in an ‘edutaining’ way (mix of education and entertainment) with emphasis on students’ own critical thinking.

From an educator’s perspective, as mentioned in Forbes (Alphonso 2023): “generative AI will better allow educators to create engaging and interactive learning experiences to help foster growth in their students” and their role will still be “invaluable and irreplaceable in the learning process”. It seems that we are already living in the future and instead of shying away from it, we should perhaps embrace it, as after all, it is here to stay.

Picture: www.shutterstock.com