Whether you realise it or not, artificial intelligence (AI) already plays a significant part in our lives.
The chances are that you use AI, or it influences you in some way, daily.
But what is AI, and how exactly does it influence our lives and everything around us?
What is AI?
In general, AI is a human intelligence simulation carried out by computer systems and is a broad term describing a range of applications and uses.
AI programming focuses on three specific cognitive skills:
- Learning Processes focus on the acquisition of data along with the creation and implementation of algorithms to interpret data and turn it into visualised and actionable information.
- Reasoning Processes also use algorithms, but the AI systems here are trained to choose the correct algorithm depending on the desired outcome.
- Self-Correction Processes continually optimise the relevant algorithms to ensure they provide accurate interpretations and presentations of data. We can also use self-correcting AI to achieve whichever other objectives we may want our system to deliver beyond data recognition.
Augmented intelligence or artificial Intelligence?
Many academic papers argue we should use the term augmented intelligence, rather than artificial intelligence.
The basis of this argument is that depictions of artificial intelligence in popular culture paint a false picture and lead to unrealistic expectations around what AI can achieve. It is debatable if we will ever reach the definition of singularity in an AI context, whereby a hypothetical artificial superintelligence agent exceeds the capability of human minds. We are likely to see the further development of AI systems, and indeed already have done, that remove human cognitive limitations such as bias, for now, the concept of superintelligence remains firmly rooted in science fiction!
Those who argue we should use the term augmented intelligence believe it's a more neutral term that will help people better understand most AI will be weak and improve processes rather than wholly replace them.
How does artificial intelligence work?
While there are several types of AI and many uses, all AI applications and systems can broadly be categorised as “weak AI” or “strong AI.”
Weak AI definition
Weak AI, sometimes called Narrow AI, refers to AI applications designed and explicitly trained to carry out a particular range of tasks.
For example, a robot working on a car production line, repeating the same process for every car that moves into its station, utilises narrow AI. Virtual assistant tools like Siri, Alexa, and Cortana are typical examples of weak AI that you might already use in your day-to-day life.
Strong AI definition
Strong AI is sometimes called artificial general intelligence (AGI). Strong AI is used to describe applications and systems that can more closely replicate human cognitive abilities. As such, a strong AI system should theoretically be able to pass the Turing Test, which tests whether an AI system is capable of thinking like a human.
True strong AI does not yet exist, although many experts believe it will become a reality in the next 25 – 30 years. When people are fearful of AI taking over human jobs and functions, it is the development of strong AI systems that is their concern.
Four types of artificial intelligence
In 2016, Michigan State University professor Arend Hintze presented an article that categorised AI into four distinct types based on their functionality.
1. Reactive machines
Reactive AI machines do not have memory functions and are programmed to complete specific tasks. The Deep Blue chess supercomputer is an example of a reactive machine in an AI context. Deep Blue was capable of making predictions and taking action based on the identification of data, in this case, pieces on the chessboard. However, it couldn't remember the steps it took, so was lacking the human ability to learn from earlier occurrences.
Subsequent chess supercomputers have been built with memory functions.
2. Limited memory
Limited memory AI systems have some memory and thus can use prior learning when making decisions in the present. Some self-driving cars use limited memory to remember things like journey times and specific features of journeys to optimise their performance.
3. Theory of mind
This type of AI is not yet widely available. However, theory of mind AI is capable of understanding emotions, therefore having the potential to interact with humans and potentially predict behaviour. Such skills will be essential for any future integration of AI systems or robots into human teams.
If you have seen the film Minority Report, where the computer identifies murderers before they commit the crime, theory of mind AI is a similar concept.
Self-awareness would give AI systems consciousness. In a similar manner to strong AI, it’s self-awareness and consciousness that many fear in a context of AI taking over human functions.
Six examples of artificial intelligence technology
As we said earlier, AI is likely to be already all around you and impacts your life far more than you probably realise. Here are six examples of AI that are already widely used.
Whether you use accounting software to help you run your business or use email marketing tools to stay in touch with your customers, AI drives all these automations and the data they gather.
2. Machine Learning
Machine learning and deep learning is essentially the automation of predictive analytics and other tasks. Credit reference agencies and comparison websites use artificial intelligence and machine learning to help make decisions around what products are suitable for you and whether they will accept your credit application.
Why do you think these platforms ask so many questions? The more data they have, the better their systems can learn and optimise how they work!
3. Machine Vision
Machine vision is used in systems required to capture visual information. Signature identification software and tools used in medical image analysis rely on machine vision to work effectively.
Some people categorise machine vision as being the AI version of human eyesight. However, this is a misguided comparison, as machine vision systems can be built to see through walls and around corners, for example.
4. Natural language processing
Natural language processing sees human language interpreted by computer programs. The spam filters for your email account rely on natural language processing alongside machine learning to identify the common indicators of junk mail.
As well as the written word, natural language processing can also be deployed in:
- Automated translation services.
- Sentiment analysis, which is used by apps like Grammarly.
- Speech recognition, which is becoming increasingly common as a means of replacing the need to remember passwords or PIN codes when calling your bank or another financial services provider.
Robots are probably the type of AI most of us are most familiar with, particularly in the example earlier of the car production line.
6. Self-driving cars
Self-driving cars, increasingly common and one thing that could potentially start to replace human workforces soon, combine several elements of AI. These systems can pilot themselves safely and avoid other vehicles and pedestrians while recognising things like traffic signals and signage.
Elements of AI are all around us.
It is probably safe to assume that, in the coming years, some jobs will be replaced by AI functions. However, until the development of strong AI, it's unlikely that, as humans, we will see our lives significantly disrupted.
In most cases, the way we interact with AI will be with weak AI applications that enhance our lives and experiences of using specific products and services. If you want superintelligence, stick to the movies!