Right now, Artificial intelligence (AI) and Machine Learning (ML) are transforming the world as we know it. We cannot escape this inevitable change. AI and ML are disrupting industries and organisations as you read this. All of us should understand the basics of this fast-evolving field and consider the implications for our own lives, our careers and the organisations where we work.
WHAT IS ARTIFICIAL INTELLIGENCE?
Artificial intelligence (AI) underpins most of what is going to happen in our new world. It’s been around for a while, and its importance will accelerate post Covid-19. Intelligent machines are influencing nearly every facet of our lives to help improve efficiencies and augment our human capabilities.
A working understanding of AI is critical for just about every career because machines can learn and make decisions much quicker and more accurately than humans. There are many types of artificial intelligence, where instead of being programmed what to think, machines can observe, analyse and learn from data and mistakes just like our human brains do.
AI is a technology that makes computers/computer programmes able to imitate the workings of the human mind. AI is active in self-driving cars, playing chess, proving theorems, and playing music.
Financial institutions, law firms, media companies and insurance companies are all working on ways to use artificial intelligence to their advantage. AI is used to discover fraud, write news stories, and review legal contracts.
The technology is altering industries from finance to manufacturing, from health care to mining with new products, processes and capabilities.
Types of AI
AI can be classified into Vertical or Horizontal AI. Let’s explain this.
A single task is performed, such as scheduling meetings or automating repetitive work, etc. Vertical AI bots perform just one job for you and do it so well, that we might mistake them for a human.
Multiple tasks are performed in horizontal AI. Cortana, Siri and Alexa are examples of Horizontal AI. These services are able to answer questions such as “What is the temperature in Nababeep?” or “Call Sipho”. They are able to perform multiple tasks, unlike vertical AI.
AI developed analysing how the human brain works when solving a problem and then using those analytical problem-solving techniques to build complex algorithms to perform similar tasks. An algorithm is a process or set of rules to be followed by a computer when doing calculations or solving problems. It’s pretty much the equivalent of a recipe to bake a cake. Except that the computer follows the set of rules every single time, without mistake. It is the set of instructions a computer will follow in order to complete a task.
AI is an automated decision-making system, which continuously learns, adapts, suggests and takes actions automatically. AI requires algorithms which are able to learn from experience. This is where Machine Learning [ML] comes into the picture.
WHAT IS MACHINE LEARNING?
Machine learning is the process of teaching a computer system to make accurate predictions using the data available to it. It is different to traditional computer software in that a human developer hasn’t written code that instructs the system what to do. The machine learns by itself! ML consists designing and applying algorithms that are able to learn from previous situations.
ML is used for credit card fraud detection, self-driving cars and face detection and recognition. It uses complex algorithms that constantly go back and forth over large data sets, analysing patterns in the data and then instructing other machines to respond to situations for which they have not been explicitly programmed. The machines learn from the history to produce reliable results.
Machine learning models are able to give accurate predictions in order to create real value for an organisation. In machine learning, prediction refers to the output of an algorithm after it has been trained on a historical dataset and it is applied to new data when forecasting the likelihood of a particular outcome, such as whether or not a customer will stop using a product or service. It helps to find the best model that represents our data and how well this model will work in the future.
Data scientists use machine learning to solve critical business problems. They extract valuable and actionable insights from big data. You have seen machine learning at work without being aware of it. Machine learning is used in internet search engines, email filters to sort out spam, websites to make personalised recommendations, banking software to detect unusual transactions, and lots of apps on our phones such as voice recognition.
No matter what industry you are now working in, machine learning is rapidly changing the way things get done. Now is the time to take courageous decisions to build your skills so you don’t get left behind.
Types of ML
There are two types of ML:
In supervised learning, training datasets are provided to the system. Supervised learning algorithms analyse the data and produce an inferred function. This function is then applied to new situations.
Supervised machine learning is -for example – where a computer is given a large number of photographs. Along with the photographs is information about what is on them. The computer analyses, sorts, and interprets what is on the photographs. This is called training. You then give the computer new photographs without the accompanying information and let it analyse, sort and interpret them, using the learning from the previous set.
Supervised learning can be used in sorting oranges to identify fruit that does not conform to a particular visual standard. It is also useful for decision support: a computer can be trained with large amounts of information about actual and historic share market transactions. It can then be used to assess how effective a intended transaction will be. Credit card fraud detection is another example of supervised learning algorithms in practice.
Unsupervised learning algorithms are more complex because the data is unsorted. The machine learns on its own without any supervision. The solution to any problem is not provided. The algorithm has to find the patterns in the data. An example of supervised learning is recommendation engines which give you a selection of options on e-commerce sites. Facebook’s friend suggestion mechanism is another example.
To put it simply, in unsupervised machine learning, you give a computer a number of photographs of each of 5 different people. The computer must interpret the visual imagery to separate them into 5 different piles, one pile for each person. Unsupervised learning is useful for clustering, and there are many applications for customer segmentation in marketing.
One industry commentator has estimated that there were fewer than 10,000 people around the world who have the necessary skills to create fully-functional machine learning systems. Machine learning engineers capable of developing these systems must be skilled in computer science, programming, mathematics, statistics, data science, deep learning, and problem solving.
Reinforcement learning is the training of machine learning models to make a sequence of decisions. The computer employs trial and error to come up with a solution to the problem. To get the machine to do what the programmer wants, the artificial intelligence gets either rewards or penalties for the actions it performs. Its goal is to maximise the total reward.
Machine learning is an increasingly integral component of solving business problems. One of the really useful ways it can help organisations is by telling us about what is likely to happen in the future.
AI and ML have reached industries like customer service, e-commerce, finance. By 2020, 85% of the customer interactions will be managed without a human.
AI AND ML CONCEPTS
Let’s explore some AI and ML concepts:
Neural networks are sets of algorithms, modelled loosely after the human brain, that is designed to recognise patterns. They interpret sensory data through a kind of machine perception. The patterns they recognise are numerical, contained in vectors, into which all real-world data, be it images, sound, text or time series, must be translated.
Neural networks help us cluster and classify data. You can think of them as a clustering and classification layer on top of the data you store and manage. They help to group unlabelled data according to similarities among the example inputs, and they classify data when they have a labelled dataset to train on. The neural network decides what the similarities and the differences are.
Each step in a neural network involves a number of guesses. A weight is a factor that either increases or decreases the significance of data inputs in machine learning, and thus speeds up the learning. But if the weight is not accurate or is set at too high or too low a setting, it will have dramatic influences on the final outcome.
In this context, bias to refer to computer systems that systematically and unfairly discriminate against certain individuals or groups of individuals in favour of others. Depending on how the algorithms and weights are set up, the system may have a bias for or against a certain outcome. This is relevant when tracking fraud or systems for awarding loans; people may be penalised or rewarded unfairly. There is a strong ethical component to the role of weights in neural networks, because they operate without human intervention.
A convolutional neural network recognises and classifies images. The algorithm guides the computer in this process. Images and pictures are particularly difficult for a computer. You and I will look at a picture and see a tree, a cow and a house. The computer sees a flat array of pixels. The computer has no sense of depth or perspective. In order for convolutional neural networks to interpret the flat pixel image, the computer compares the image with a very large set of image libraries. It goes back and forth between the libraries and the image. Hence the term convolutional. The algorithms initially look for and compare looking simple features such as edges and curves, and then build up to ‘see’ more abstract and complex concepts. When completed, the computer ‘sees’ the house or the cow as you and I would identify it.
This is called visualisation. It brings a visual dimension (2 or 3D) to an object that is described in non-visual data. Visualisation allows an object to been “seen” without physically being there, with the help of these algorithms and convolutional neural networks.
Deep learning is related to this. It’s a type of machine learning where artificial neural networks learn from large amounts of data. They do this by themselves once the algorithms and neural networks are set up. In other words, the ‘machine’ can go ahead (provided it has enough data or images) and solve complex problems by itself. That is super cool!
At its simplest, deep learning can be thought of as a way to automate predictive analytics. While traditional machine learning algorithms are linear, deep learning algorithms are stacked in a hierarchy of increasing complexity and abstraction.
Applications in practice
A chatbot is an artificial intelligence (AI) software that can simulate a conversation with a human being in natural language through messaging applications, websites, mobile apps or through the telephone. A chatbot is often described as one of the most advanced and promising expressions of interaction between humans and machines.
Machine learning is being transforming advertising. Programmatic advertising is the automated process of buying and selling advertisements through an exchange, connecting advertisers to publishers. Programmatic advertising uses machine learning technology to improve efficiency and make better budget decisions for advertisers.
Real-time bidding is one of those techniques. It refers to the buying and selling of online ad impressions through real-time auctions that occur in the time it takes a webpage to load. As the ad impression loads in the user’s browser, information about the page and the user is passed to an ad exchange, which auctions it off to the advertiser willing to pay the highest price for it. The winning bidder’s ad is then loaded into the webpage nearly instantly; the whole process takes just milliseconds to complete.
Another technique is a recommendation engine. It is a system that suggests products, services, information to users based on analysis of their data. The recommendation is based on factors such as the history of the user and the behaviour of similar users.
Here are some common applications which use AI and ML:
Automatic language translation:
Machine learning can convert text in one language into another. It can also assist with pronunciation.
Email spam and malware filtering
Machine learning algorithms used for email spam filtering and malware detection.
Image recognition is a common application of machine learning. It is used to identify objects, persons, places, digital images, etc. The best example is the Facebook automatic friend tagging suggestion.
Machine learning is used to quickly and accurately diagnose disease. As a result, medical technology is underdoing revolutionary change.
Online fraud detection
Machine learning is very effective in detecting fraudulent transactions. The algorithm makes our online transactions more secure by detecting aberrant transactions.
Machine learning is used by e-commerce and entertainment companies such as Amazon and Netflix, to recommend products to the user. Our previous search histories are used to come up with a set of recommendations.
The application of ML in self-driving cars is still in its infancy. Machine learning plays a significant role in self-driving cars. Unsupervised learning is used to train the car control systems to detect people and objects while driving.
Share market trading
Machine learning is used extensively in share market trading. The ML neural network is used to predict share market trends.
Speech recognition converts voice instructions into text, and it is also known as “speech to text.“ Examples are Google Assistant, Siri, Cortana and Alexa.
If we have a smart phone, we are familiar with apps like Google Maps and Waze. They work out the shortest route and predict the traffic conditions along the way.
Virtual personal assistant
Virtual personal assistants such as Google Assistant, Alexa, Cortana, and Siri help us to find information using our voice instruction.
Each industrial revolution has brought with it new ways of how we work and what we produce. The 4IR, the one we are living through now, is unmatched because of the sheer acceleration of change. Artificial intelligence and machine learning are driving this acceleration.
IMPLICATIONS OF AI AND ML
What are the implications of AI and ML?
More and more jobs will become automated. Jobs that require repetitive action, or quickly sourcing and interpreting data will be automated. To find out how automation will impact jobs, click here
The way we work will also change. Because boring reportative work is automated, our roles in jobs will change. We will have to be more creative and people-oriented. There will be a greater demand for problem-solving, empathy, listening, communication, interpretation, and collaboration. – those skills where humans are better than machines. In other words, the jobs of the future will focus more and more on the human element and soft skills
This is how Deloitte sees new categories of work emerging:
Deloitte sees the emergence of super jobs. These new types of jobs, combine into one, roles which up to now have been largely separate: manager, designer, architect, and analyst. These are roles that combine work and responsibilities from multiple traditional jobs, where technology is used to both augment and widen the scope of the work, involving a more complex combination of technical and human skills.
Dramatic changes in how the workplace functions are under way. More and more people are working from home. Robots and cobots are no longer strange. Tools will support meetings and flag work processes that are falling behind. The psychological contract of employment is reshaping before our eyes.
AI and ML are going to change our lives without a doubt – the process has already begun. A number of big tech giants including Google, IBM, Microsoft, Facebook and Amazon, created the Partnership on AI to research and advocate for ethical implementations of AI and to set guidelines for future research and deployment of robots and AI. Find out more about the partnership on AI, click here.
WHAT NEXT FOR ME?
If you want to make a career out of AI, you will need some fundamental knowledge of a computer language such as C, C++, Java, Python, etc. You will also requires competency in mathematical concepts such as derivatives and probability theory.
On the other hand, you may want to understand more about AI and Ml. You want to be able to manage people who work in AI and ML, and you want to be able to hold your own in meetings about AI and ML. You want to actively be part of the 4IR and just be a bystander.
Regenesys has a range of digital courses, at the Basic and Advanced level that will have you talking like a pro in no time at all. The basic concepts and process are explained, step by step. And the beauty of it is that it is all online. You need never leave the comfort of your home or your office. With a Regenesys Certificate, you will put yourself ahead of the pack.
Take that critical first step. Click here to find out more about getting ahead in the 4IR.