Generative AI

Some generative AI is trained on vast amounts of text data and deeply understands language, including grammar and vocabulary. This means that it can help students better understand the complexities of language and provide them with instant, accurate answers to questions.

AI can provide students with real-time feedback on their writing, helping them to identify areas for improvement and develop better writing skills over time.

AI can make education more accessible, as students can access information and support from the model at any time, from anywhere with an internet connection (providing the technology is fully functional). AI can also be used to generate content in different formats, such as audio or visual aids, making it more accessible to students with disabilities. This can ensure that all students have an equal opportunity to learn and succeed.

AI chatbots, such as Copilot, Gemini & ChatGPT, can provide a more interactive and engaging learning experience, as students can interact with the model in real-time and receive personalized responses to their questions. The software is trained on a massive dataset of text, which means it has access to vast amounts of knowledge. It can provide answers to questions on a wide range of topics and can help students discover new ideas and perspectives. Additionally, the conversational tone can help to make learning more accessible and less intimidating, especially for students who may struggle with more traditional, passive learning methods. 

Generative AI can be used to generate new ideas and research questions for academic research, providing a starting point for further exploration and investigation.  

Generative AI’s natural language understanding capabilities could be used to grade student essays and other written assignments, potentially providing more accurate and efficient scoring than human graders. 

AI tools such as Copilot, Gemini & ChatGPT could be used to generate multiple-choice or short-answer test questions, which could save time and effort for educators. 

Other opportunities include virtual tutoring, quick question answering, providing personalised feedback on student work, and more. 

Large language models generate text from online sources. They are trained on massive datasets of text which can contain biases and inaccuracies. So, they are more likely to follow trends which are wide spoken but may contain outdated or incorrect information. This means that particular biases or misconceptions can be reinforced, such as historic gender biases that exist within numerous academic fields, like the prioritisation of male symptoms within health subjects. Inaccurate or incomplete responses can be especially problematic in fields that require a high degree of precision, such as the sciences.

AI cannot comprehend the emotion behind both what it is being asked to say or do. This means it does not understand the context of the conversations it is having. While ChatGPT can simulate human conversation, it cannot replace the value of human interaction. Some students may find it difficult to learn in an environment that lacks the personal touch of a teacher or mentor. 

As with any technology, there are ethical considerations to using ChatGPT in higher education. Institutions must consider the potential impact on student learning and the potential for bias in the system’s responses.

Additionally, the reliance on generative AI technologies could lead to a reduction in critical thinking and problem-solving skills as well as a lack of creativity in the learning process – all of which are essential for academic success. This is because educators and students may become overly dependent on such tools as automated essays or responses. People may trust that the AI is correct rather than doing their own research on the topic area. AI cannot think critically, insert references, or develop that breadth of knowledge personal to users (e.g., students and their essays). By using AI in a way which might undermine students’ skills, they will not be achieving their learning outcomes, so it may stunt their academic development.

ChatGPT does not provide any explanation for the result it produces. It just returns the text. It is not useful to understand the reasoning behind the results or to evaluate the quality of the arguments.

Most AI is currently unable to distinguish between fake and real information and could generate false citations and statistics, affecting the quality of work produced. This could be more obvious when asking the Chatbot to explain a complex or niche topic. Or even something that is not widely available. 

Most applications of AI are still limited and have a narrow scope in terms of outputs and real-life applications (limited algorithms).

AI can produce text that is usually grammatically correct and coherent. However, it cannot critically or creatively do this. This implies that it may struggle with creating original ideas required for academic work.  

Example

Generative AI tools may give you a lot of cliches for your work or overused themes or theories. This is because the AI will have limited knowledge to pull from its database and will not be able to expand on that knowledge creatively.

Although certain generative AI systems may have been trained on a substantial amount of text data from numerous sources, some languages are still left out. As a result, it might not be able to comprehend or produce text in those languages, which can be a major drawback for students working in multilingual settings.  s more challenging.

AI may struggle because of a lack of context. Although it can create text depending on the information supplied, it may not completely comprehend the context or background of the study question or issue. This might result in erroneous or partial replies that are unhelpful to researchers. Moreover, ChatGPT may lack access to specific expertise or resources required for academic research, limiting its effectiveness in this setting. 

Example

If you ask an AI chatbot a question on the impact of climate change on the economy. It may generate responses based on the patterns in its training data. It may not consider specific details such as the economic sector or the geographical location of the impact.

AI relies on technology, which can be prone to malfunctions and downtime. If the system experiences technical difficulties, users may be unable to access relevant (learning) resources or get the help they need when time is of the essence.

ChatGPT uses large amounts of data to generate its responses. This raises concerns about data privacy and security. Institutions must take measures to ensure that student data is protected and that ChatGPT is not used inappropriately.