.png)
Understanding Artificial Intelligence
Learn a little about how AI works and gain a fuller understanding of the limitations of Generative AI technology.
What is Artificial Intelligence?
Artificial Intelligence, or AI, refers to computer systems that can perform tasks that usually require human intelligence, such as writing text, generating images, producing music, recognising speech, translating language, and making informed decisions. AI is based on the idea of creating machines that can "think" and "learn" like humans do, using algorithms to process and make sense of large amounts of data.
AI has the potential to revolutionise the way we access education, learn, and work, and we need to know enough about it to be able to use it responsibly. But what exactly is AI?
Sorry, you need JavaScript to view this video
Narrow AI versus General AI
AI tools can be divided into two very broad categories: Narrow AI and General AI. We probably all use Narrow AI without realising it. This type of AI is limited to specific, automated and repetitive tasks such as online recommendation systems and voice, image and text recognition.
You will find Narrow AI features in lots of University digital tools. In Library Search, for example, Narrow AI is how the system suggests keywords as you type into the search box, and recommends books and journal articles you may like to read. Microsoft's AI features such as Designer in PowerPoint and Editor in Word are both examples of Narrow AI - these systems suggest ways to improve your work, but only know what they are programmed to do.
General AI, on the other hand, is a type of machine intelligence that is more like human intelligence. It is able to learn, adapt and connect knowledge from many different sources to solve problems. This type of AI can reason and adapt… just like we do.
Generative AI
Generative AI exploded onto the scene for many of us following the public launch of ChatGPT in 2022. This AI technology is able to generate new outputs based on a set of training data, using what is called a Large Language Model (a deep learning model that has been trained on large amounts of data and understands and generates text in a human-like fashion). Unlike other AI tools, Generative AI creates new and original content such as images, text, audio, video, and code.
Generative AI is a technology that is more understandable and accessible for many of us. By inputting a few simple prompts you can quickly get started using it.
How does it work?
The core of a Generative AI is a trained deep-learning model (or software program) that understands and generates text, images, or other media in a human-like fashion based on a given user input, i.e. a prompt. This model is trained on massive amounts of data to learn from patterns in the dataset. For example, it will learn that certain words tend to follow others, or that particular phrases are more common in certain contexts. The model then uses the prompt to produce a completion, or output, which is presented back to users.
The quality of the generated output will depend on a number of factors, such as the amount and quality of the training data, the complexity of the prompt, and the model's size. Larger models usually generate better outputs but require more computing power and resources. For example, ChatGPT, Copilot and Gemini focus on language generation while Midjourney and DALL.E generate images.
AI for Learning
Enrol on this short Canvas course to learn more about how AI works, the limitations and challenges around critical use of outputs, and try out prompting for yourself.
What are the limitations?
Even though AI generated content is generally well presented and appears convincing, the tools can, and often do, get things wrong. You should always question the output, apply your judgment concerning its reliability, and fact check the information provided. Many AI tools are unable to reference their sources and you will find that citations are often fabricated.
AI is only able to generate responses based on the information it is trained on and the available dataset:
- the data is not always current and tools may be drawing on sources that are months or years out of date
- if the dataset is biased this bias will be transferred into the generated content. Without careful analysis, biases, stereotypes and in some cases Western perspectives may be perpetuated.
AI performs better when it has more sources to draw on. Tools are likely to produce comprehensive outputs on subjects that are widely written on, so if you are looking for assistance with more specialised or niche areas, or cutting edge research, the information may not be so well developed.
AI tools can not apply human critical thinking or the development of an evidenced argument:
- it can synthesise information on a debate but can't assess which side has more strength or credibility
- it may not pick up on subtleties and nuances in the writing, which the writers use to express their stance
- AI tools can't apply the knowledge generated to the real world, except in a very superficial sense
The skills that AI lacks are vital for analysing, developing, and expressing an argument. Critical analysis is a very human skill and often something that is built into the assessments you are tasked with at university. Application of knowledge to differing contexts is also a fundamental skill and your ability to do this helps you to develop higher order thinking skills.