GenAI V2

Overview

Generative AI technologies are changing the higher education landscape in a deep and profound way. As the integration of these technologies becomes increasingly prevalent in educational settings, it is imperative for both staff and students to engage with these tools in a responsible and ethical manner This guide serves as a comprehensive resource to facilitate this learning journey.

We begin by introducing Generative AI, explaining what it is, and highlighting its significance in higher education. The guide then delves into the core components of Generative AI, including its applications and potential impact across various disciplines.

A central focus of this guide is the development of AI literacy. We provide insights into the basics of AI and machine learning, and emphasize the importance of critical thinking in the context of AI. Practical exercises and resources are included to enhance understanding and application.

Ethical considerations form the backbone of responsible AI engagement. The guide explores the ethical dimensions of AI, including biases, fairness, privacy, and intellectual property. We provide best practices and case studies to illustrate responsible AI use, and offer guidance on developing institutional AI ethics policies.

Academic integrity is another focal point, as we address the challenges posed by AI in assessments and academic work. Strategies for designing AI-resilient assessments and encouraging originality are discussed, along with tools for detecting AI-generated content.

Looking ahead, the guide examines the evolving nature of higher education and the role of AI in shaping the future of work. We discuss the adaptation of curricula and teaching methods, and the importance of fostering a culture of lifelong learning in the AI era.

In conclusion, the guide summarizes key takeaways and encourages continued ethical and responsible engagement with AI. A rich repository of resources for further reading is provided, along with a glossary of AI terms and sample AI ethics policies.

Whether you are an educator, student, or institutional leader, this guide is an invaluable resource for navigating the complex landscape of Generative AI in higher education with responsibility and ethical integrity.

What is Generative AI?

Generative AI is a type of artificial intelligence that enables users to generate novel outputs resembling human-created content, including text, images, videos, sounds and 3D models. Recent breakthroughs in the field have drastically improved the capabilities of these models, making generative AI the fastest adopted technology in history.

The crux of generative AI is in its name – generative. While traditional AI algorithms have focused on identifying patterns within data for predictive purposes, say to predict whether the next image in a sequence of images is a cat or a dog, generative AI leverages the learned patterns to create entirely new outputs. Given the above example, a generative AI model could create an entirely new species of cat, or dog, or cat-dog!

How does Generative AI work? (SH)

Generative AI is grounded in machine learning techniques that draw inspiration from the neural systems of the human brain, known as neural networks. These neural networks are ‘trained’ on extremely large amounts of data, during which they learn to capture and internalise the essential features, patterns and relationships within the data. This enables the model to generate new data instances that are similar yet distinct from the original training data.

For instance, a generative AI model trained on images of faces learns to understand facial features like nose shape, eye placement, and smile curvature, allowing it to generate new faces that possess the same structure and features but do not match any specific face. Similarly, models like ChatGPT, trained on vast amounts of language data, learn to understand grammar, sentence structure, common phrases, context, tone and style, enabling them to generate coherent and contextually appropriate text.

Generative AI can be run on a variety of models, each employing unique methods for training the AI and generating results. These include generative adversarial networks (GANs), Variational AutoEncoders (VAEs) and denoising diffusion probabilistic models (DDPMs).

Why are Students Using Generative AI?

Understanding why students might decide to use AI for their coursework is critical to ensuring that we genuinely support learning, and that the institution meets its obligation to graduate employable and ethical citizens. Students might elect to use genAI for their coursework for numerous reasons, and it is important to understand that not all of these reasons are mischievous or with an intent to cheat. Some students may lack confidence in producing work entirely themselves, whilst others may not feel motivated or supported to do so. Indeed, scholarship on why students participate in academic dishonesty more widely suggests that the possible reasons can extend beyond the desire to achieve certain results to include: feeling inadequately prepared for assessments; caring more about results than learning; confusion around what constitutes academically dishonest behaviour; feeling like the behaviour is commonplace amongst their peers; or feeling a lack of connection to their studies or institution more generally (Bryzgornia, 2022). Some scholars have even raised the notion of ‘ethical cheating’ in reference to students collaborating, sharing knowledge/information/ideas and using open-source platforms precisely to develop 21^st century skills, yet in ways that might traditionally have been considered cheating (see Brimble, 2016).

In relation to AI specifically, a survey of US higher education students conducted by Best Colleges in March 2023 showed that students hold diverse views towards the use of AI in university coursework, ranging from those who actively use it to those who believe it should be prohibited in educational settings (Richards, 2023). In the same survey, 40 percent of respondents said the use of AI defeats the purpose of education, and 63 percent said AI cannot replace human intelligence or creativity. As discussed in the Assessment section of this guide, students deserve clarity around what is considered proper versus improper use of AI in their studies and for each assessment task. This includes when and how students are meant to disclose the use of AI tools, and any distinctions around expectations when it comes to AI use in text-based versus graphic-based formats.

As Siva Vaidhyanathan writes, “We have been dealing with cheating methods and technologies as long as we have been asking students to prove their knowledge to us.” Apart from clarifying university policies and expectations, it may be beneficial to discuss with students the use of AI by professionals and academics in the field, and the current set of ethical questions surrounding these practices. The key is to maintain an open dialogue with students about the use of AI in terms of what is permitted or restricted, versus what is encouraged or even required. Not only are tools and technologies certain to develop over time, institutional and personal stances towards AI are context-dependent. Generally, if students feel uncomfortable or discouraged to discuss their views or habits with staff, this can contribute to a problematic gap between teacher assumptions/expectations and learner practices. ­­The more educators and students can feel like they are working together to promote learning and professional development the better. As Ouyang and Jiao (2021) argue, the advancement of AI technologies does not ensure good education outcomes; rather, the long-term goal of AI use in educational contexts is to contribute to a paradigm where learners are supported and empowered to take agency for their own learning.

Datasets and Bias

Paradigms of AI usage in Higher Education: According to Ouyang and Jiao (2021) there are three paradigms of how AI is being utilised in education. AI-directed, where learner is considered as recipient (paradigm 1), AI-supported where learner is perceived as a collaborator (paradigm 2), and AI-empowered, where learner contributes as a leader (paradigm 3). Paradigms 1 and 2 have been the focus of AI education in the past two decades. However, there is a need for Paradigm 3, i.e., AI-empowered, Learner-as-Leader to be enacted in Higher Education that is centred upon promoting human intelligence and AI to be integrated, so that issues of biases in AI algorithms and datasets, lack of governance and how an AI system makes decisions and offers outputs for learning and teaching experiences can be more socially just and inclusive.

Types of Datasets: For learning and teaching experiences to be socially just and inclusive regarding the use of AI, it is important to understand how students use the AI platforms and what forms of information do they input or are provided with as an output. This involves use of datasets that are text, image, audio and/or code based. In BE education, students can engage with AI platforms using either of these types of datasets depending on the subject type and thematic topic/content under focus for their learning.

Following is a basic description of different types of datasets:

Image Datasets: Generative AI heavily relies on image datasets to train and generate new visual content. Images are typically classified, detected, and segmented using image datasets that consist of large collections of labelled or unlabelled images. Generative AI models are developed for image synthesis, style transfer, and other creative purposes with the help of these datasets.

Text Datasets: Generative AI text datasets are an essential component of natural language processing (NLP) models. These datasets are collections of text data that are used to train artificial intelligence models to generate coherent and meaningful language.

Audio Datasets: Audio datasets are used to train generative AI models. These datasets are used to generate audio content such as music and audio synthesis. Datasets include collections of audio recordings, including single sounds and full-length songs, which are used to train machine learning models to produce new, original audio.

Tabular Datasets: From financial analysis to predictive modelling, tabular datasets are frequently used to train generative models. In tabular data, data imputation is a common application of generative models.

Datasets can be grouped together and may lead to several outcomes depending on whether data in one form leads to an output on an AI platform in another form such as shown in the diagram below (this is a simplified version)

https://www.cogitotech.com/generative-ai/

Depending on the type of the AI platform such as ChatGPT, the text datasets may not be carefully curated or selected for the AI platform to produce outputs relevant to cultural, linguistic, ethnic, and historical backgrounds of the topic for learning or the socio-cultural background of the student.

Data Sets usage/role along the paradigms: Datasets in Gen AI can be used for textual outputs such as essays, reports, summaries, reflective narratives, thesis, rendered images, development of audio outputs and drafts. Gen AI can be used for data analysis including learning analytics and data from LMS.

Types of biases/exclusion points in Datasets: Algorithms in AI platforms and systems/Machine Learning systems that seek to do things better by increasing efficiencies can embed existing biases and replicate existing conditions of disparities. Large social datasets feed systemic bias into algorithms, and unchecked algorithms can result in systemic discrimination that favours certain individuals or groups over others. Machines rely on training through use of large amounts of datasets in different forms such as text, images or audios to recognise and produce outputs in response to prompts entered or the nature of task demanded. Most datasets are Western-centric because of the dominance of these forms of information that are readily available for the Gen AI platforms such as ChatGPT to utilise. This can hinder the achievement of social justice in classroom and decolonisation efforts in place by higher educational institutions.

"Although AI can assist in creating substantial and insightful content, it is not a simple task and demands proficiency in both technical skills and knowledge of ethical considerations. Students need to be made aware of, for instance, stereotypical ChatGPT answers. As ChatGPT is trained on a large dataset of text, it is very likely that the data used to train ChatGPT may contain societal biases (Dahmen et al., 2023). Consequently, the model will reflect these biases in the output data and, thus, reinforce existing societal issues and discrimination (Atlas, 2023). Therefore, when using ChatGPT in higher education, students should be aware of this potential bias and critically reflect on each statement created by ChatGPT (Gimpel et al, 2023). " 

Ferrara 2023,

When asked ChatGPT, ‘ChatGPT do you use datasets that project exclusion and promote bias?’, its response was as follows:

As an AI language model, I don't have direct access to information or knowledge beyond what I have been trained on. I have been trained on a mixture of licensed data, data created by human trainers, and publicly available data. My purpose is to assist and provide information to the best of my abilities, regardless of biases that may be present in the training data. However, it's important to note that I do not have control over the training process or the selection of the data I was trained on.

OpenAI, the organization behind ChatGPT, is committed to addressing biases and promoting fairness in AI systems. They have guidelines in place to prevent the use of datasets that explicitly promote bias or exclusion. They strive to provide a system that is useful and respectful to users from various backgrounds.

If you come across any biases or concerns in the responses I provide, please let me know, and I will do my best to address them.

Role of the ‘what’ of datasets in relation to the ‘how’ and ‘why’ of AI literacy:

Datasets have a lifecycle of input, usage and interpretation. It is important that at each stage of the lifecycle, the students are supported in how they relate to the data and its interpretation.

(limitations and implications of current datasets for diverse learners,  i.e., [capabilities (physical and neurological)/transdisciplinary (varied skills and competencies)/international (cultural, linguistic, ethnic, religious, racial diversities) and how teachers can utilise culturally responsive learning and teaching methods – (section to be added with further research)]

Challenges to Creativity and IP Rights

Some questions related to Creativity and Authorship around genAI are especially relevant to Built Environments disciplines and learning.  As outlined above, genAI tools draw on and learn through data sets.  They search, re-combine, and deliver elements from data sets in response to requests from a user producing a wide range of outputs in text, numeric, code and graphic form, and more.  They also collect vast repositories of requests and data from users, directly or through ‘plug ins’, to ‘learn’ and refine their data sets for future requests.

When we ask students to ‘be creative’ in design-related disciplines or learning activities, we are asking them to bring their own beliefs, values and attitudes to the generation and judgement (iteration) of infinite potential resolutions to a design challenge. Students are often asked to identify and respond to relevant precedents, research and experiences for their design problem.  They learn to select from and/or transform these ideas, as well as to reflect on and direct these practices as they develop their expertise.  We ask them to participate through this in the ‘curious and beautiful relation between design problems and their solutions’ (Lawson (2007); Lawson & Dorst (2009); Cross et al (1994)).  Students and practitioners are rewarded for innovative responses that move beyond such precedents and familiar solutions, and contribute positively and innovatively to our world.

Many doubt the capacity for AI to participate in this ‘true creativity’ (Lawson (2007); Kelly (2019)), as it lacks the motivation and independent judgement to create the truly new and useful. Human thinking processes are described as creative and flexible by nature, in contrast to the strengths of AI in relation to repetitive actions at vast scale, managing complexity and multi-tasking (Wu et al (2021)). Some claim the Turing Test (in which an outcome that is indistinguishable from human production offers proof of intelligence) has been reached, while others claim creativity should be relocated to the perception of the viewer, rather than the contribution of a potential author (Natale & Henrickson (2022)). Others agitate on the part of the machines for genAI as its own form of authorship and complain that the ‘goalposts’ of creativity are being moved beyond their reach (Arielli & Manovich (2022)).  It is important not to confuse the ‘output’ with an assumption about the maker – a key challenge at the heart of the assessment challenge in an era of ChatGPT.

In other areas, purported opportunities to ‘collaborate’ with GenAI see iterative models of design thinking transformed to show human and computer contributions to identified phases. These outline perceived improvements to human thinking / expressing / building / testing / perceiving, by increasing scope and decreasing time and cost through human-AI co-creation (Wu et al (2021)); or the ways in which different disciplines may creatively understand and engage with AI (Wingstrom et al (2022)).

Elsewhere, claims of human authorship have been debated and legislated for some time, but human and AI production is being considered through the lenses of moral rights (Miernicki & Ng (2020)), intellectual property and copyright (Shtefan (2021)). A recent case before the US Copyright Office has found the location of authorship via input (prompts) vs output (AI produced images), although the debate continues.

Simultaneously, the developers of various GenAI platforms are in legal hot water, as artists claim these companies are infringing copyright by drawing on the published work of artists without attribution, and the opportunities for recourse or even protection are surely limited (see a summary at https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem ). When AI is drawing on its ‘own’ previous output for the raw materials of future production, it all gets a lot more complicated.

These deliberations offer useful prompts for our own consideration of student originality in their study, and effective education for changing professional roles.

Designers and others are using genAI in numerous ways to support creative professional practice (Wingstrom et al (2022)). As an example, Zaha Hadid Architects is using AI to inform and drive personalisation of office spaces (https://www.nytimes.com/2023/06/15/business/workplace-design-zhai-ai.html ) within a design project by increasing collection and fine-grained analysis of data. Elsewhere in the firm, AI text-to-image generators are being used to come up with design ideas for projects, and for early ideation (https://www.dezeen.com/2023/04/26/zaha-hadid-architects-patrik-schumacher-ai-dalle-midjourney/). The use of ‘Zaha Hadid’ in the prompting is reportedly used by the studio principal to claim authorship and inclusion of GenAI outputs in the oeuvre of the firm.

A GenAI focussed panel discussion at the recent CSHE Teaching and Learning Conference heard panel members encouraging the design of learning experiences in which students could refine their creative practices and deepen their judgement by ‘sparring’ with the machine through reflective use of text prompts – in these approaches, as elsewhere, it is important to remember that these tools work by delivering and linking statistically likely elements.  It is also important to remember they are drawing on these prompts, as well as our judgement of the outcomes, to refine the algorithms. We need to ask where our students’ judgements and values will come from, and how they will test and refine these in the context of our disciplines and elsewhere.

Our challenges include supporting students to engage creatively with the tools to hand, to build their own creative expertise, and to effectively demonstrate authorship and to protect their own IP and privacy in the process. In these spaces the notion of creativity and its expression remain central, and a crucial aspect of the learning that students need space and support to practice and refine.

Generative AI and Assessment

>> Assessment/learning design recommendations : SS (and all … this is the outcome informed by the above)

The role of assessment tasks in education is to facilitate evidencing a student’s learning. Increase access to generative AI poses complex challenges for educators, particularly in regards to designing meaningful assessment tasks that accurately capture a student’s learning.

Colluding Collaborator or Learning Partner: Key Considerations in GenAI Assessment Design

In assessment design, AI affords students to cognitively offload while they perform tasks to evidence their learning. Cognitive offloading is the use of additional tasks to find other pathways for information processing to reduce cognitive demand and strain (Dawson, 2020). For example, when we jot down items on a grocery list so that we avoid forgetting things to purchase we are cognitively offloading by reducing the need to cognitively remember what needs to be picked up at the store. AI platforms afford students opportunity to cognitively offload and engage in higher levels of cognitive processes such as critical thinking, reflective thinking, problem solving and creative thinking.

When considering the design of their subject(s) assessment tasks and integration of AI the following questions should be considered:

• What ILO’s is the subject(s) designed to teach and develop? Is there opportunity to align/update the ILO’s with AI literacy skills as identify in Future/s of work with AI?
• How can students demonstrate their learning if they are collaborating with an artificial buddy? Are there tasks that can be designed specifically for students to engage in higher levels of cognitive thinking?

(Anatomy of Chuck’s Brain):

Cope et al. (2021) state that AI can provide the four functional parameters to a learner. The following table outlines these four functions that

Cope, B., Kalantzis, M., & Searsmith, D. (2021). Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies. Educational Philosophy and Theory, 53(12), 1229-1245.https://doi.org/10.1080/00131857.2020.1728732

https://staff.flinders.edu.au/learning-teaching/good-practice-guides/good-practice-guide---designing-assessment-for-artificial-intell

Considerations for assessment designs to engage students’ capacity to evaluate and create:

• Centre the assessment task on a context and/or situation where AI cannot access information and data on. E.g. assess materials discussed in class as a summary of tutorial discussions.
• Shift the focus of what is being examined/assessed from the final output to the “behind-the-scene” process. E.g. the assessment task may instruct students to keep a detailed reflective/design journal that documents the process founding their final outcome
• Collect evidence of learning through modes that AI technologies are unable to replicate/output. E.g. students may participate in synchronous conversations / interviews

Considerations for assessment design to collaborate with AI

AI and Academic Integrity