Exploring Generative AI: critical skills and ethical use

Welcome to this Open University Library recording of "Exploring Generative AI-- critical skills and ethical use." I'm Carol Wright, and I'm a learning and teaching librarian at the Open University. In this session, we're going to explore generative artificial intelligence, or AI, covering what it is, how it works, and we'll also provide guidance and tips to help you ensure that your use of Generative AI is responsible, ethical, transparent, and aligns with the OU guidance. We will be focusing on the skills needed to use AI tools, rather than how to use the tools themselves.

Learning outcomes. By the end of this recording, you will be able to recognise the strengths and limitations of generative AI to make ethical and responsible decisions when using AI tools, critically evaluate AI tools and their outputs to assess how appropriate and reliable they are for your needs, create effective prompts for generative AI tools to generate relevant and accurate outputs, and appropriately acknowledge the use of Generative AI in your work following University guidance.

Artificial intelligence, or AI, in everyday life. Before we start, have a think about what ways might you already interact with AI in everyday life. Pause the video and have a think about this. Examples of AI in everyday life. Lots of us have been interacting with AI in various ways for years, sometimes without realising it. Some of the examples we thought of were chatbots used in customer service, shopping, banking, booking a holiday, and tech support.

Facial recognition-- unlocking your phone and in camera apps. Digital assistants, for example, Amazon Alexa, Google Assistant, and Siri. Spell check and autocorrect. Travel and navigation, for example Google Maps. Writing assistants, for example Grammarly. Health and fitness apps, for example Fitbit. You may also have come across AI assistants in library databases. And you may also be using generative AI for things like meal planning or interior design ideas.

But what is generative AI? It is a new type of AI technology that automatically generates or creates new content such as text, images, audio, video, and code in response to prompts given to it by users. ChatGPT and Microsoft Copilot are examples of text-generative AI tools, while Adobe Firefly is an AI tool that generates images, video, and audio. But how does generative AI actually work?

Generative AI tools are trained on vast amounts of data from sources like web pages, social media conversations, and other online content. They learn to recognise patterns in the data, such as what word or phrase is likely to come next, and use this to create new content when given a prompt. For example, text-based generative AI tools like Microsoft Copilot and ChatGPT are built on large language models, or LLMs, and work like predictive text, although a more sophisticated and large scale version. And they can continuously improve by collecting and learning from our previous questions and prompts. Each response is unique. You will probably never get the exact same reply twice.

Limitations and risks of generative AI. Using generative AI has some limitations and risks. Sometimes generative AI will produce information that is incorrect. We call these hallucinations. Generative AI tools present information in a convincing way, so it's very easy to be misled by incorrect information. GenAI can display bias or stereotypes. It may generate content which is stereotypical, racist, or sexist. GenAI usually gives very generic answers to questions, as it lacks the emotional intelligence and critical thinking skills that you have yourself.

GenAI has limited understanding of context, and this can only be gained by giving the tool context by using effective prompts, which we will talk about later. Creating GenAI tools and using them has an environmental impact. There is also a human cost to the development of AI tools, including the use of low paid and an often invisible labour and the emotional toil of managing harmful or distressing content.

Sustainability of generative AI tools. As we already mentioned, GenAI has an environmental impact because training and running models require large amounts of electricity, water for cooling data centres, and computing hardware, all of which contribute to carbon emissions and resource depletion. To reduce these risks, consider whether you really need to use AI for this particular task. Are you using the most efficient tool available for this task? Could a simpler approach, for example a Google search or using Library Search achieve the same results?

Activity-- comparing AI and human-generated text. One way to gain a better understanding of the limitations of AI is to compare AI and human-generated content. So we asked a human and a generative AI tool, in this case, we used ChatGPT to perform the same task. We asked them to critically analyse and reflect on how you conducted your literature review. I'm going to share the responses in just a moment, and I'd like you to spend a few minutes considering them. Can you identify which response was written by a human? You might like to pause the recording to consider them.

Activity responses. Response A was written by a human. ChatGPT provides a general description of how a literature review might be undertaken. Although it uses "I," which is misleading, it is still quite a vague statement. It doesn't give anything really specific. It's factual. The AI-generated text also feels quite impersonal, even though it is talking in the first person. It also uses American English spelling of the word "synthesising" and "analysing." ChatGPT defaults to American English because a large portion of its training data comes from US-based websites.

The human answer includes factual information, but also elements of reflection, analysing different types of the topic. This is a very human skill. A generative AI tool like ChatGPT cannot reflect on its own thoughts or apply critical thinking to develop an argument in the same way that a human can. The human response is much more specific and gives examples.

For example, it names Library Search as a tool which would provide access to more key studies and names the evaluation tool Prompt. This activity has shown us that generative AI should not be seen as a replacement for things like writing, critical thinking, and evaluation. These are essential skills that you will need to develop throughout your studies and in everyday life.

Open University guidance on generative AI for students. The OU has developed guidance to help you to use these technologies responsibly and effectively in your studies. You can use generative AI to support your learning if you want to but you must check your module guidance to see if generative AI use is permitted for assessment. The OU guidance provides examples of how these tools can be used, along with information about which tools you can use.

It includes guidance on how to use generative AI in an ethical and responsible way, and there is a section on generative AI and assessment which covers how to check if your module permits the use of GenAI for assessments, and how to reference and acknowledge the use of generative AI tools. A link to the Open University guidance is available on the YouTube page and also on the Open University Library Training and Events page. You will need an Open University login to be able to access this. The guidance will be reviewed annually to ensure it reflects developments in generative AI in education, and we'd encourage you to check it regularly to stay informed about the latest information and advice on using generative AI for your studies.

Generative AI for learning. Keeping the challenges and limitations of generative AI in mind, there are ways that you can use it to support your learning. You could use it to obtain alternative explanations, to generate quizzes to test your knowledge, to critique your work, to check for spelling errors and clarity, to practise your skills, for example, your language skills, and to support a literature search by suggesting keywords. Library Search and some library databases also include built in AI tools to help you search, summarise, and analyse the information you find.

Using generative AI ethically. Generative AI tools can be effective tools for learning, but to ensure they are actually enhancing learning, they need to be used in an ethical and responsible way. Always protect privacy and confidentiality. You must not provide confidential or personal information about any individual or organisation, including yourself and the OU, to any generative AI tool, including Microsoft Copilot in protected mode. Always check the terms of use and privacy policies of any AI tools you use to understand how your data will be used.

Always respect copyright. Do not provide AI tools with any copyrighted material by the OU or any other individual or organisation, including journal articles, unless you are using Microsoft Copilot in protected mode or have been given permission by the copyright holder, or your module has instructed you to do so, or you are using a tool provided to you as a reasonable adjustment for your disability.

Using AI responsibly. Check your module assessment information for guidance specific to your module and also the Open University AI guidance. Be transparent in your use of Generative AI. Give credit to the source of any AI-generated content in your work by referencing it, as you would any other source. Also, be aware of what might be considered plagiarism. Do not copy and paste from AI tools and claim it as your own work.

As we've already discussed, generative AI tools can sometimes generate information that is incorrect. Always verify generated information using your prior knowledge of a subject, or by using reliable sources such as your module materials or Library Search, library databases or Google Scholar. Assessment and feedback are essential in developing transferable skills, such as communication, research, and critical thinking. If you overly rely on generative AI, you run the risk of not developing these skills.

Effective prompts. When you ask a generative AI tool to perform a task for you, the instructions you give it are called prompts. Developing your prompt-writing skills will enable you to use the tools more effectively and efficiently. I mentioned the importance of effective prompts earlier. The AI can only work with the information you give it, so its response will only be as good as your input. So how can you make sure your prompts are effective? I'm going to talk through some tips for effective prompt writing, followed by a demo to show some of them in action.

Clearly state what you want the AI to do or to generate. Specify the format or structure you want the response in. For example, do you want a paragraph or a list or a summary? Give the AI relevant context to understand the task better. Giving the tool a role is a great way of providing context. For example, you could ask it to act as if you are a university tutor in history. "I am a final year in history studying at a UK University."

Write your prompts in natural conversational language. Don't worry about having to think like a machine. Ask your question as if you're having a natural conversation with the tool. Specify what you do and don't want the tool to provide. For example, you want it to provide a list.

Build on your previous prompts by carrying on the conversation. You don't need to start again. If the task is complex, break it down into smaller subtasks or steps. Ask the AI to tackle each step individually. Be patient. Be prepared to change or refine your prompts, as you may not always get the perfect response first time.

I will now demonstrate creating an effective prompt. For this exercise, I'm going to use Microsoft Copilot. There are many different generative AI tools available, but Microsoft Copilot is the only safe tool from a copyright perspective for OU study. It ensures that data stays within the OU and is not used by the tool for training and that Microsoft won't claim copyright on the generated outputs. You can access Copilot in various ways, such as using a direct link for it or in the Edge browser.

Make sure you log in with your ou.ac.uk account, and check that there is a shield in the top corner. This will ensure that enterprise data protection applies. Remember, even when using Copilot, you must not input personal data. For our demo, I want to show how thinking about your questions and prompts when using AI can make a difference.

For this example, I've decided I want to improve my diet and eat more leafy greens and fruit high in vitamin C. I want some help with meal planning, some meals for the week. I'm going to use the prompt "meal plan leafy greens" in Copilot, a bit like we would with a Google search and see what it comes up with.

So it's given me a seven-day meal plan, and it's given me some ideas of how I could incorporate leafy greens into my meal planning. But it hasn't been very specific. It doesn't tell me what the ingredients are for these meals. It doesn't tell me how much of the ingredients I need, and it doesn't tell me how to make them.

It's also suggested some ingredients that I don't like. For example, I don't like fish. Now, Copilot has done a good job at responding, but I haven't been specific enough for the tool to deliver what I really wanted. So I think the first thing I need to tell it is to exclude fish. So you don't need to start a new prompt. You can just carry on the conversation.

[KEYS CLICKING]

 

So please exclude fish. So by doing this, we've immediately solved one problem with the meal plan. And it now excludes ingredients that I don't eat. However, I don't really have enough information here to follow the plan in practise. Here's an example of a better prompt for this. So the prompt is, "I want to eat more leafy greens and fruits with vitamin C. You are a professional nutritionist skilled at planning healthy diets and meals. Provide a set of specific recommendations for how I can do this. Make sure you provide enough variety so that I can rotate my meals. Provide exact ingredients, quantities, and instructions."

 

So you can see that the results displayed would allow you to make these meals. If you didn't know what any particular ingredient was, you could ask Copilot to explain or to do an internet search. So it's given me different examples for things that I could have for breakfast, for lunch, and for dinner, with instructions on how to make them and the amounts of ingredients needed.

So some of the reasons that the results have been more helpful for me are because in the prompts, I have given Copilot a role as a nutritionist to provide context. I am being clear and precise. I was able to use natural language. I have reviewed and refined my request based on the previous results, so for example, by excluding fish. I learned from additional searching that unless I specify, it won't give me quantities.

Using generative AI can be an iterative process. You keep improving and refining your requests until the tool delivers what you are asking it for. What you may find is your initial request is more complex than you imagined. Some tools limit how much information they deliver to you in one entry, and so you may need to break down your request into smaller chunks. For example, to ensure I get all the information I need for my seven-day meal plan, I may need to do each day separately.

How to use generative AI critically. You will need to use your critical thinking skills throughout your studies. Whether it's critically evaluating journal articles or a website, these skills are particularly valuable when using generative AI. Not only do you need to critically evaluate the output created by the AI, but you need to evaluate the tools themselves.

Think before using AI. Do you need to use AI for this task? Again, think about the environmental impact. For example, if you're planning to create an image, do you need to use a Gen AI tool to create a new image? Or is there a high quality image from one of the libraries copyright cleared databases that already exists and you could use? Think about the tool you want to use. Check its strengths, limitations, environmental impact, privacy policy. Do you know anything about the data it was trained on? Is this the correct tool for your purpose?

Check facts, verify sources, and check any references. Use trusted and reliable sources such as Library Search, library databases, or Google Scholar to check every output that you create. When using library databases, you may come across AI tools which are built into the database. You need to use these critically, the same as with any AI tool. Consider the usefulness, reliability, and ethics of these features.

Review any outputs for bias or stereotypes. Ask follow-up questions to address them, or use a different tool to address this. Look at diverse viewpoints through different sources, as AI may not provide a diverse range of viewpoints on a subject.

There are a number of different places to go to for help. You can use the OU guidance on Generative AI for students. You can check your module assessment information and ask your tutor. You can also contact your Student Support Team, or if you're a postgraduate research student, the Graduate School. For help with referencing generative AI, you can consult the Cite them Right Harvard guidance for Generative AI. You can also contact the Library Helpdesk-- contact details are on every page of the Library website.

Additional activities. There are a number of different additional Library activities on generative AI that you can do. These activities are listed on the YouTube page and also on the OU Library Training and Events page for this recording.

Learning outcomes-- revisited. You will now be able to recognise the limitations of generative AI, critically evaluate generative AI tools and their outputs, create effective prompts for generative AI tools, and appropriately acknowledge the use of Generative AI in your work. Thank you for watching.