What is AI and how do I use it?

1. What is Artificial Intelligence?

Different people can mean different things when they use the term Artificial Intelligence (AI). 

One common definition is that Artificial Intelligence is the name we give to the study of making computers perform tasks that would normally need a human’s intelligence. 

The set of technologies that perform these tasks are known as “AI technologies”.

Newer, often more helpful, definitions of Artificial Intelligence tend to avoid the comparison with “human intelligence” and focus on them being machine-based systems that create output such as content or a decision based on inferences they make from data.

2. Why is it hard to define “Artificial Intelligence”?

It can be hard to agree on a definition of “AI” for many reasons. 

AI technologies are also used in many different ways by many different people, and the technologies can work in very different ways, so what counts as “AI” can be up for debate – even amongst computer scientists. 

3. How do AI technologies work?

AI technologies are based around three building blocks: data, algorithms and models.

Data is information that is stored on computers in a useful way. This can be any type of information, such as letters, numbers and images. AI technologies require very large collections of data.

Algorithms are the instructions given to computers to let them make predictions based on data. 

An AI model is the arrangement of algorithms to perform useful tasks.

To create and improve AI technologies, these models are “trained” on big amounts of data. How well the AI technology performs depends on the quality of the data it is trained on.

4. What are some examples of “AI technologies”?

One very famous example of an AI technology is ChatGPT. 

This AI technology is known as a “Large Language Model” (LLM) and can respond to questions you ask, generate text and produce content.

LLM’s answers depend on the data it has been trained on, and the quality of the answers depends on the quality of the data it has been trained on. ChatGPT has been trained on a huge amount of data from a vast range of sources, which means it’s able to respond to a huge amount of questions. 

There are also voice recognition AI technologies, which can recognise what people say when they speak out loud using a technique known as “Natural Language Processing”. This is what allows smart speakers like Amazon Alexa to be able to respond to questions. 

Artificial Intelligence techniques, such as “Machine Learning” are also used in social media and on the internet to personalise content, present you a tailored newsfeed and to target ads.

These technologies are good at recognising patterns, and are used to flag spam email or to protect people from fraud in online banking.

AI technologies are used to help make decisions, such as whether someone is eligible for a certain treatment in a healthcare setting. Or to flag when a health condition needs looked at by a doctor.

AI is used in many, many different ways across many different sectors: for wildlife conservation, land use mapping, crime prevention, stock management, data analytics and in the creative arts. People are using these technologies for new things every day.

5. How might disabled people use AI technologies?

Some AI technologies are used to help people communicate through speech.

People with different speech patterns can use an app they train to recognise the way they speak, displaying it in words and allowing for easier communication. This enables users to speak in their own voice while making communication easier when interacting with others.

‘The Voice’ app is one example. It helps people with impaired speech communicate using voice and control voice-driven smart speakers. Users train phrases and commands, and the app learns to recognise their pronunciation. When it hears a trained prompt, the app speaks the matching phrase and displays it on screen.

Some AI technologies are used to help blind people and people with visual impairments.

‘Seeing AI’ is an app from Microsoft. With the use of image recognition technology and a smartphone’s camera, the app will “speak” aloud what it is seeing to that person. Users will be alerted if, for example, there is a table in their way, or that a picture has a dog in it, or that they are in a kitchen.

6. How can AI help older people?

Generative AI can be useful to help with drafting letters and emails or to help with research topics and areas of interest. This can help older people save time in contacting services, or getting information they need. It can also create personalised learning experiences to help with learning something new, and you can ask them to explain something to you in exactly the way you want.

ALISS, a service in Scotland, uses an AI search assistant called ‘Heather’ to help people find services, groups and activities for health and wellbeing. It is trained with all the information about what’s available in the local community, and you can chat with it to find out what would best suit your needs.

AI-powered devices can help people monitor their health, as they are very good at dealing with patterns. This can alert the user or those who support them to any signs of health issues if the patterns change.

Smart speakers are powered by AI technologies, and can be helpful with managing schedules. This could help people who take medications at a regular time, or keep track of refills. 

Many projects within AI aim to help older people live independently for longer. For instance, AI systems are being trialled that monitor people’s gait in their household, to allow for early support intervention before a fall. This can reduce the need for a fall monitor.

7. Can you trust AI for accuracy?

The accuracy of AI tools and technologies depends on how well it was made and how well it was trained, so it depends on each use of AI. AI technologies used in healthcare, for instance, undergo very stringent training and checking, and are only deployed when their accuracy is assured. 

There is a saying within the world of data and AI, “garbage in, garbage out”. This means that the quality of the tasks completed by an AI system are dependent on the quality of the data it was trained on. This is why transparency around AI is useful, as it means that users or a trusted organisation can check to see how well the AI technology performs.

With tools like ChatGPT that produce text, it is useful to know that it is a system that is trained to output the most likely answer based on the data it has been trained on. This means it is not an oracle of trust, and you should always sense-check and fact-check anything you are producing using it. 

8. What are the risks of AI technologies?

There are many different risks around AI technologies that we need to consider to take full advantage of AI.

Bias, discrimination and exclusion are risks of using AI technologies. AI needs to be trained with data, and if that data is biased, discriminatory or exclusionary then the outputs of the AI technologies could reinforce the inequalities and barriers that exist in the world.

Our human rights can be impacted when AI technologies are used to help with decision-making in important areas like immigration, policing, housing, recruitment and welfare. If AI is used inappropriately or has been trained on biased data then it could violate people’s rights and privacy.

When AI technologies are used to automate work, this can change the types of jobs that are available or how people are expected to work. This means that people could lose employment opportunities.

The use of algorithms by social media sites and websites can lead to a number of online harms. For example the promotion of extreme views or content can lead to problems with mental health and hate crimes.

9. Are there laws or regulations around using AI technologies?

There are not specific laws around the use of AI at the moment, but in the UK they must be compliant with the General Data Protection Regulation (GDPR) and the Equality Act 2010. 

Different organisations that use AI may have their own rules and regulations around using AI.

Scotland has a national AI Strategy, which has a vision for Scotland to become a leader in trustworthy, ethical and inclusive AI.
The EU has recently passed an EU AI Act that regulates AI based on the risk of its impact on people.

10. How can AI be trustworthy?

AI can be trustworthy if it is transparent, so that we can observe how these systems help organisations make decisions, and must be disclosed so that people are aware of the use of AI in their lives.

Decisions made by trustworthy AI must be explainable, and we must have confidence in the reasoning behind them.

When AI is trustworthy, we can be assured that it will function securely and safely and that its risks are continually monitored and managed, and are protected from attacks or misuse. 

11. How can AI be ethical?

AI can be ethical if it is used for good, and respects and supports our values. AI that supports our ambitions to be a fairer and greener society while enabling us to become a more outward-looking and prosperous country.

It should support our human rights, our environment and our communities, while being accountable to the people who are affected by it and uphold our laws and our rights.

12. How can AI be inclusive?

AI can be inclusive if the conversations around AI include everyone from across the diverse cultures and communities that make up society.

Working to include groups who may be particularly affected by AI is important to have inclusive AI, particularly children and young people, and under-represented or marginalised people. 

The technologies should be shaped by a diverse range of voices in all areas from development, design, implementation and use and support us to live free from discrimination.

13. What are AI technologies good at?

AI techniques and technologies tend to be very good at recognising patterns and learning from data. This means they can automate tasks that we as humans may find repetitive or boring, freeing us up to concentrate on things we enjoy more.

AI may be used in healthcare to speed up processes or make processes more accurate, which will allow doctors to concentrate more on dealing with patients. There are numerous trials underway around the use of AI to detect and flag health issues, based on being trained on the patterns in health data. 

Tools known as Generative AI are good at producing content such as text and images. Generative AI tools have become quite advanced over the past 18 months, creating convincing text and pictures.

ChatGPT is one of these tools, and it can help you structure a letter or an email, or help you get started on a piece of writing. You can also use it to assess your own writing, by giving it a task such as “Assess this writing for style, impact and structure – and give me 3 tips to improve it”. 

AI technologies work best when you’re using them as part of a human and AI team, rather than just setting the AI off on a task and not checking how well it works. The more that there are humans in the loop, the more powerful the use of AI will be. 

Businesses can take advantage of AI by automating processes behind the scenes. This allows for streamlining and efficiency, that in turn is creating more productivity. This, combined with progress in other emerging technologies such as robotics and data science, has lead to some people calling the rise of AI technologies a new “industrial revolution.” 

We are at the very start of how we work with AI, and if we can ensure that our relationship with these technologies is grounded in trust, ethics and inclusion, then the advantages can be shared by everyone in society. 

With many thanks to the Scottish AI Alliance.