General
What is AI?
There’s a lot of buzz around artificial intelligence, also referred to as AI. While it can seem quite alarming, especially when you hear news reports about the risks it poses to jobs, our privacy and society in general, it also comes with many opportunities. There are ways AI can help our lives, and you may even be surprised to hear you may already be using it without realising.
In this article we provide a beginner’s guide to artificial intelligence and break down what it is, examples of how AI is being used today, some of the ways it can help us and some things to look out for.
In this article:
- What is AI?
- How does AI learn?
- Examples of AI
- Different types of AI
- What are some of the ways AI can be used to help us?
- Downside to AI
What is AI?
AI is an umbrella term that refers to machines that can perform tasks the way humans would do. Or to be more technical, AI is an engineered system that produces content, forecasts, recommendations or decisions in response to a set of human-defined tasks without programming.
Machine learning is a term sometimes used synonymously with AI, however it’s not the same. Machine learning is a type of AI that uses algorithms (or sets of instructions) to train computers to recognise patterns from enormous amounts of data to make a prediction or decision. Machine learning accounts for most of today’s AI.
How does AI learn?
AI learns from being fed large amounts of data such as text, images, speech and algorithms. The data and algorithms combined allow AI to complete a task.
Take for example ‘training’ AI to produce a drawing of a cat. To do this, AI software reviews millions of pictures of cats so that it recognises key features and patterns that allow it to learn how to create its own version of a cat. If it gets it wrong and produces an image of a dog instead, it learns from its mistake by adjusting the algorithm to get the desired output.
Examples of AI
You are probably already interacting with AI in your day-to-day life without realising. Here are a few common ways AI is currently being used:
Recommended content on subscription services
If you’ve watched Netflix or listened to Spotify, you may have noticed they come up with suggestions for you on what to watch or listen to. That’s AI and it’s called the recommender system.
Steve Nouri, a data science leader and the founder of AI4Diversity explains, “the recommender system is an AI algorithm that understands our watching patterns and engagement patterns and will come up with a complex understanding of what our brain likes and what makes us interested in watching more.”
The algorithm recommends what it thinks we will like, and sometimes even suggests different things that it thinks we may be interested in.
Voice assistants
Voice assistants like Siri and Alexa use voice recognition which is a form of AI technology. When you ask Siri or Alexa to check the weather, the machine learning algorithm understands and responds to your voice. You can find out more about voice assistants and how they can help you with day-to-day tasks in our short course on smart home technology.
Chatbots
Chances are you’ve interacted with a chatbot when you’ve visited a website – it’s the small pop-up box that often appears in the bottom right corner of a web page that offers to help answer questions you may have. When you type in a question for the chatbot, it analyses what you’ve entered and tries to understand what it means by comparing it to pre-defined questions and conversations that it’s been programmed to recognise.
Facial or image recognition
The later smartphone and tablet models offer facial recognition as an alternative option to unlocking your device with a PIN. Your device maps your facial features from an image you’ve taken then compares them with a database of faces to find a match to unlock your device.
Search engines
Search engines like Google and Bing are using AI to improve your search experience. Normally when you do a search online, you’re presented with a list of related organic links and paid ads so that you can discover the answer for yourself. Search engines that use AI may provide a summary of information found online as an answer to your search query. This gives you the option of not having to read through individual websites.
Different types of AI
The most common form of AI that exists today is narrow AI. It’s designed to focus on a defined task and is used to address a specific problem. It cannot perform tasks outside of what it’s been programmed to do. For example, if a machine has been programmed or ‘trained’ to produce a drawing of a cat, it wouldn’t be able to draw any other animal.
One of the advantages of narrow AI is the speed and efficiency with which it can perform tasks compared with humans. However, the downside is its inability to understand the wider context of the task it performs. For example, online language translations such as Google Translate use AI to translate words, however they don’t consider the meaning or nuances behind everyday language so can miss things like humour, sarcasm and emotion, which can affect a translation.
By contrast, general purpose AI is a type of AI system that has a broad range of tasks and uses. Generative AI (gen AI) is a subset of general-purpose AI and uses machine learning to generate content such as text, video or images, or a combination of these, in response to a prompt.
Some examples of gen AI applications include ChatGPT, a popular text-based chatbot that can be used to simulate conversations with humans, write essays, poems, or even CVs to name just a few. DALL-E is another popular gen AI tool used to generate images instead of text.
There are other forms of AI including artificial general intelligence and artificial superintelligence that aim to think, feel and learn like a human. However, these are still hypothetical concepts.
What are some of the ways AI can be used to help us?
AI is transforming the way some industries operate by improving efficiency and automating repetitive tasks. For example, banks and other financial organisations are using AI for fraud detection to analyse and recognise suspicious transactions that could signal fraud.
AI can also be applied in healthcare to reduce human error and help make faster and more accurate diagnosis. For example, it’s being used to make more accurate predictions of alzheimer’s disease and completing repetitive admin tasks leaving clinicians more time to work directly with patients.
Downside to AI
AI can be used to help our day-to-day lives, however it also poses some risks.
Generative AI tools can be used to create fake images, videos or voices of real people – also known as deepfakes - that spread fake news, disinformation, and scams. Deepfakes use footage or pictures of real people to create convincing and entirely fake photos or videos of them doing or saying something they didn’t do or say.
Deepfakes can be used in scams where celebrities or public figures appear to promote a product or service which ultimately is fake. You can find out more about fake celebrity endorsements on the Scamwatch website .
While generative AI tools such as chatbots or text generators can be helpful, they can also provide inaccurate or outdated information so it’s best not to exclusively rely on them as a source of truth, even when they sound believable.
Also, as gen AI tools learn from the data used to train them, they can produce information that’s biased, even if it’s unintentional. Watch the What is AI? video by the Museum of Science for real world examples on how misconception and bias can creep into AI.
For more information on AI
- Listen to the Be Connected podcast episode, 'Artificial intelligence explained' with guest Steve Nouri to find out more about the benefits and risks of the technology.
- Watch 'AI Explained: It’s maths, not magic' video by the CSIRO.
- Read articles on AI on The Conversation .
- Read 'eSafety’s position statement on Generative AI' including opportunities, risks and advice for users.
- Read 'eSafety’s position on Recommender systems' including their benefits, risks, and impacts.