malvernmugadzikwa
Dr Dennis Magaya
ARTIFICIAL INTELLIGENCE (AI) is poised to have the most profound impact in human lives in the history of technology in the 21st century.
AI refers to computer systems that mimic general intelligence found in human beings.
This allows people to use and interact with technology as if it could hear, see, feel and talk like humans, which possibilities and opportunities that were not feasible before.
AI usage is already widespread in our daily lives.
For instance, when we open WhatsApp, there is a Meta AI blue circle on the phone screen, and we use ChatGPT to give us information on many things.
The use of AI for social economic development requires a basic appreciation of what this technology is all about, starting with the devices that we know and use daily.
Laptops, mobile phones, calculators and other computer-based technologies have a huge number of instructions which are executed depending on the command the person types on the keyboard.
Traffic lights have inbuilt instructions to sequentially switch from green, amber and red.
These are early foundations of AI, where the intention is to give a computer instructions so that it works on behalf of humans.
Computer technology continued to improve and more intelligence was built in such that the first breakthrough to have AI was in 1956.
Since then, four broad categories of AI emerged.
The design philosophy behind all of them is for computers to emulate humans through automation and learning.
Human beings apply their knowledge and experience daily to solve problems and make decisions.
This human characteristic leads to the basic AI category called Symbolic AI or expert knowledge systems.
The idea is to translate the knowledge and experience of experts into a programme so that computers can be used for analysis and decision-making instead of humans.
The computer programmes are constructed using a set of “if … then … else” rules.
For instance, “IF the salary is $1000 AND the age is 35 years, THEN the loan is US$50”.
“IF you see a flame AND feel high temperature AND smell smoke, THEN it’s a fire.”
As the experts get more knowledge, the AI system is enhanced with more rules.
The advantages are that millions of rules can be programmed so that decisions are made quickly.
Secondly, the computer isn’t forgetful, doesn’t go to sleep, doesn’t take leave like humans.
The expert system can be automated to work 24/7.
Some self-service chatbots and Interactive Voice Recorded (IVR) systems for call centres are Symbolic AI systems.
When the chatbot fails to address the query, it means the request scenario is not recorded in the rules and so the system escalates to a human being.
The disadvantage of rules-based systems like Symbolic AI is that they are difficult to use in vague, complex, dynamic and multivariable situations.
There are also scenarios such as colours and emotions that can’t be translated to rules. So, there was need to develop a better AI which is as close as possible to human beings.
For instance, a child knows about fire when the parents teach them that when you see this flame, smell this smoke, feel this heat and hear the cracking sound, then it’s a fire.
Therefore, a computer programme can also know a fire provided it is given data and trained that whenever it sees this picture or video, get this temperature range, sees this type of smoke and hears this sound, it is called a fire.
Such a computer is equipped with sensors to detect images and sound and measure temperature.
This AI technology is called machine learning (ML).
It means the computer has a programme that can be trained to analyse and make decisions based on the presented data.
However, just like a human being, the computer (machine) must learn first and its accuracy improves from training on more data.
The ML technology can also be programmed to learn on its own just like humans.
Imagine how children surprise parents after learning something on their own; that’s what AI does too.
The computer programmes used for machine learning simulate the brain neurons, hence the computer algorithms are referred to as neural networks.
Computers work on data and, therefore, as long as data is available, ML can be used to mimic human intelligence in many applications.
Since computing power literally doubles every two years, ML can process huge amounts of data in milliseconds.
The data can be video, audios, pictures, text, PDF or sign language.
Humans generally communicate with computers using requests typed on keyboards.
These are processed and a response is provided on the screen.
Unfortunately, the keyboard can’t capture everything, including emotions and tone of voice which we would want to communicate.
In addition, a person is better understood if they use their mother language compared to when they use another language.
Therefore, AI gives even better results if it can understand human’s natural language.
The AI technology called Natural Language Processing (NPL) enables computer systems to understand the natural languages so that people can talk to computers instead of using static menus.
The computer programme should be trained first so that it learns the language.
Just as children are taught to read and pronounce the alphabet, verbs, nouns, adjectives, pronounces, computers are also taught the same.
They also learn different pronunciations, accents and language dialects.
In addition, they also learn how to interpret emotions and sentiment from the voice tone.
ChatGPT uses natural language processing to understand your request and goes into a large database to retrieve your request.
The other AI technology is called computer vision, which is the use of computers to recognise, understand and classify images or videos of objects or people.
The combination of symbolic AI, ML, NPL and image processing technologies are used to develop AI software programmes that solve complex problems just as humans would do.
Although AI cannot completely replace a human being, however, there are some tasks that can be completely done by AI, for example, data capturing and loan processing.
Home appliances, cars and advanced systems such as satellites and weapons use AI.
It has found use in the medical field where cancer can be diagnosed and operations performed by robots.
Using AI for diseases diagnosis is much easier when huge amounts data should be processed to extract patterns and make predictions.
In agriculture, AI can be used to detect crop diseases, determine the required moisture for irrigation and fertiliser.
When farmer simply sends a picture or video or audio describing the crop, the AI is able process and give a response.
In education, AI can be used to provide personalised learning and teaching.
In business, there are several AI use cases.
As long there is data, any problem can be solved.
Feedback: [email protected] / +263 71 777 0666
Online Reporter