KARACHI:
What do people do to increase the accuracy and efficiency of their work? They get the information that their brain needs to react, compute, and carry out a task. What actions do computers take to improve their own effectiveness or accuracy? They employ a process known as machine learning (ML). Companies are utilising artificial intelligence (AI) as the globe develops and they work to increase their productivity in an effective manner (AI). With the aid of great brains, Pakistan, which is also making progress in the IT sector, is working to keep up with the rest of the world in the field of AI and ML. Even yet, the majority of them are employed by foreign firms because of Pakistan’s resource shortage.
Although AI is prevalent now, there was a period when the area as a whole wasn’t seen as significant. After early successes and much hoopla in the mid-to-late 1950s and 1960s, scientific advancements stopped and fell short of expectations. Insufficient computational power prevented that potential from being realised. Additionally, the cost of running such a system was exorbitant. As a result, money and interest stopped flowing. The 1980s saw a resurgence of this endeavour because to greater research funding and a growing algorithmic toolbox. However, the decade-long AI winter quickly returned.
Then, two crucial developments happened, directly paving the way for modern AI. Rule-based systems were the focus of early AI research; however, ML approaches that can learn from data without external programming are being used. ML depends on the explosion of data and data sharing that has occurred as a result of the World Wide Web becoming commonplace in the hands of billions of people worldwide.
Artificial intelligence (AI) is the capacity of computers or computer-controlled robots to carry out operations requiring human intelligence and discretion. An AI system based on the learning process of human neural networks is used by Siri and Google Translate as examples of AI use cases.
Many say that AI will improve the quality of our daily lives by being more efficient than humans at basic and complex activities, making life easier, safer and more efficient. Others argue that AI threatens people’s privacy, classifying people to exacerbate racism, displacing employees and increasing unemployment.
Machine learning
A branch of AI known as machine learning (ML) enables software programmes to anticipate events more correctly without having to be explicitly programmed. ML algorithms forecast new output values using past data as input. In the previous 10 years, we have advanced significantly.
Although AI and ML are sometimes used interchangeably, they differ significantly. The word “AI” refers to a group of technologies that allow computers to learn and behave like people. AI, in essence, creates the
In ML or augmented analytics, input data and output are supplied to an algorithm to generate a programme, as opposed to conventional programming, which consists of a handwritten programme that accepts input data, runs it on a computer, and creates output. This produces significant learnings that may be applied to forecasting future events.
In order to uncover patterns in voluminous data, including words, figures, and pictures, machine learning algorithms employ statistics. Data may be entered into ML algorithms to tackle certain issues if it can be stored digitally.
According to ML Engineer Abdul Raheem, the first approach was developed from his pure statistics in the 1950s, The Express Tribune said. “They found patterns in numbers, assessed how close data points were to one another, and used these techniques to answer formal arithmetic problems.
“Neural networks are highly favored by major IT businesses. Of all, 2% accuracy is a $2 billion gain in revenue for them, but while they’re little, it doesn’t make sense. I’ve heard tales of teams that spent a year developing new recommendation algorithms for e-commerce sites before realizing that search engines account for 99% of all traffic. Because of their worthless algorithm, most people didn’t even access the home page “Raheem continued by saying that the sole purpose of machine learning is to forecast outcomes based on incoming data. If a job is not expressed in this fashion, there was never an ML issue.
Types of ML
There are several ways to train ML algorithms, each with advantages and disadvantages. We must first look at the kind of data that each sort of ML ingests in order to comprehend its advantages and disadvantages. Unlabeled data and labelled data are the two forms of data used in ML. Although labelled data includes input and output parameters in fully machine-readable patterns, labelling the data in the first place takes a significant amount of human labour. Data that is not labelled has one or no machine-readable parameters. Although there is no human labour involved, a more difficult solution is needed. There are other ML algorithms that may be applied to specific situations, but the three basic approaches utilised today are reinforcement learning, unsupervised learning, and supervised learning.
A pupil with an instructor is an example of supervised learning. One of the most fundamental kinds of machine learning involves labelling data to explicitly tell computers what patterns to search for. Although appropriate labelling of the data is required, supervised learning is appealing and produces outstanding results when applied in the right situation.
“By pressing “play” on a YouTube video, we’re telling the machine learning system to discover similar videos that match our preferences. Then, as the following suggested video, they are displayed “ML engineer turned software engineer Raheem explained.
Unsupervised learning, on the other hand, involves a learner without a tutor and no data labels. Because this type does not require human interaction to make the records machine-readable, the machine searches for patterns at random. This implies that you can work with considerably bigger data sets using programming. Unsupervised machine learning services are less common than supervised learning since they are used less often in daily life.
“Unsupervised learning involves giving a machine a stack of pictures and giving it the job of recognising objects in the pictures. There is no tutor, and the computer looks for patterns on its own, Raheem added. Machines are utilized more frequently for jobs that need real-world application since, of course, they pick up new information more quickly with an instructor. These include the prediction of object types via classification and the prediction of particular points along a numerical axis through regression.
Reinforcement learning primarily refers to a class of ML issues where agents function in settings without a predetermined training data set. Agents need to understand how to manage criticism. This ML algorithm suppresses unfavorable outcomes while reinforcing or promoting favorable outcomes.
Use in daily life
ML has a wide range of uses, including internal applications that assist businesses enhance their goods or expedite labor-intensive procedures, as well as external (consumer-facing) uses like product suggestions, customer support, and demand forecasting.
ML algorithms are frequently applied in fields where post-deployment improvements to solutions are necessary. Companies from many sectors employ highly dynamic customizable ML solutions.
We will witness the use of ML everywhere if we look about us. There are many uses for ML nowadays. The recommendation engine that drives social media news feeds is arguably one of the most well-known applications of ML.
ML is used in social media to customize how each user’s feed is presented. The recommendation engine will place that group’s activity earlier in the feed if members read posts in that group less frequently.
The engine works behind the scenes to reinforce recognised patterns of users’ online behavior. In the upcoming weeks, if members alter their behavior and cease reading postings from this group, their news feed will change to reflect this.
Similar applications can be found in medicine. One AI-powered technology that assists clinicians in identifying high-risk patients before they are given a diagnosis of the illness is called Deep Patient. According to inside BIGDATA, the system predicts roughly 80 ailments a patient may experience up to a year in advance based on their medical history.
In contrast, PathAI’s ML algorithms assist pathologists in properly diagnosing patients by analysing tissue samples. The quality of both the diagnosis and the therapy has increased. The algorithms used by PathAI can also choose qualified volunteers for clinical studies.
AI is used by virtual assistants like Ideal to aid in the hiring process. This new HR system aims to do away with or drastically cut down on time-consuming operations like reviewing resumes, choosing candidates, and other repetitive chores. The software utilised for HR screening makes use of data and predictive analytics. As a consequence, it improves HR department efficiency and uses strong recruitment techniques to dispel preconceptions. By translating data from multiple sources into more accurate and precise insights, AI helps with decision-making.
ML powers every piece of AI software and platform. They can’t function as well as they do without it. The system will produce better and more accurate findings as more data is entered into it.
The most frequent applications of machine learning (ML) are in the following areas: identifying spam, making product recommendations, customer segmentation, image and video recognition, detecting fraudulent transactions, demand forecasting, virtual personal assistants, sentiment analysis, and customer service automation.
Scope of ML in Pakistan
ML powers every piece of AI software and platform. They can’t function as well as they do without it. The system will produce better and more accurate findings as more data is entered into it.
The most frequent applications of machine learning (ML) are in the following areas: identifying spam, making product recommendations, customer segmentation, image and video recognition, detecting fraudulent transactions, demand forecasting, virtual personal assistants, sentiment analysis, and customer service automation.
Sabih Sheikh, a data scientist and computer science graduate, claims that AI and ML are fundamental to how the world functions, advancing many businesses and enhancing quality of life. “AI and ML have a very broad application. Pakistani AI engineers are being hired by global IT behemoths like Google and Microsoft in addition to the domestic industry. IA firms in Pakistan that pay well welcome talented IA specialists. Additionally, experts in this industry might make more money working as independent contractors. Numerous businesses in Pakistan have implemented AI and ML. However, given the limited amount of data, “He continued by saying that initiatives like PIAIC will expand the use of AI in Pakistan.
The use of artificial intelligence (AI) in medicine is expanding globally, yet underdeveloped nations like Pakistan lag behind in implementing AI-based healthcare solutions. Diagnoses, monitoring, and resource allocation can all be aided by AI. According to surveys, the majority of doctors and medical students in Pakistan are uninformed about artificial intelligence (AI) and its applications, but they have favourable views of it and are open to its adoption.
Another industry that actively uses AI and ML is journalism. By tweaking flexible paywalls for subscribers, ML can assist news companies in improving their business models. ML is already being used to improve journalists’ abilities at every stage of the newsgathering process.
Investigative journalists can and have tried incredible analytical approaches to make sense of these large datasets as data is kept in ever-greater quantities. In the process, it may hold businesses and governments responsible. ML is used for this, which enhances data-driven reporting. This method is useful and essential in the big data era.
The guideline about when to apply ML in reporting is rather straightforward. It’s time to start the machine when stakeholders can’t meaningfully examine the data manually (we’re talking hundreds of thousands of rows in a spreadsheet).
Institutes for ML in Pakistan
To incorporate young people in the development of the nation’s economy, AI centres have been created all around the country. The National Centre of AI at NUST Islamabad, the AI Research Lab (AIRL) at UET Lahore, the Center of Intelligent Systems and Networks at UET Peshawar, and the AI Lab at IBA Karachi are a few of the top AI laboratories. All of these institutions offer bachelor’s in AI (BSAI) degrees.
The Presidential Initiative for AI & Computing is one factor that will revolutionise the Pakistani industry for AI and ML (PIAIC). With the goal of transforming business, research, and education via the use of cutting-edge technology, PIAIC was established. This is what experts refer to as the fourth industrial revolution. It intends to turn Pakistan into a hub for augmented reality, data science, blockchain, cloud-native computing, edge computing, and the internet of things.
The project, which was started by the President in partnership with the Sailani International Welfare Trust (SWIT), aims to empower adolescents by providing them with advanced online course training. The Presidential Initiative for AI and Computing (PIAIC) 2022 entrance test was held at the National Stadium in Karachi with over 25,000 students from all across Sindh taking part.
“What a pleasure it has been to motivate Pakistan’s youth. 24,833 appeared in a test to start their mentored online course in Software, AI, Blockchain, IoT, Networking & Cloud computing. No other place could accommodate them but National Stadium. Karachiites will lead Pakistan,” said President Arif Alvi after the first admission test.
This program is backed by the Digital Skill Program, an online platform for IT education. In contrast, Kamyab Jawan Program extends interest-free financial support of up to Rs. 1,000,000 to youth to launch their journey to become an entrepreneur.
Career opportunities in Pakistan
There are a lot of Pakistani ML and AI specialists working there, but very few of them work for businesses that are Pakistan-based. The majority of professionals work for businesses that offer services to international businesses. “Although Pakistan has brilliant minds in this subject, the country struggles to compile data, making it difficult for individuals with degrees and experience to use their abilities for Pakistan. The absence of usable data, with the exception of a few large firms, is the reason why AI and ML are not being used, “Sabih added.
He continued by saying that there are numerous employment prospects in robotics, computer vision, language processing, gaming, expert systems, speech recognition, and many other fields for AI and ML grads and experts. He said, “Anyone interested in this career and ready to put in the time and effort to get a high level of education might easily be recruited into this rapidly expanding and demanding field.
In addition to those with a degree in the relevant discipline, those who are familiar with computers and probability can also succeed in this sector.
Future of ML
In the area of AI, machine learning is a fantastic technique. Even with its early uses, machine learning is already enhancing our present and future.
The variety of ML use cases and applications expands as ML develops. It is important to think about how ML applications may be utilised across all business domains to save costs, boost productivity, and enhance user experience in order to effectively handle the business issues of this new decade.
Efficiency may increase if two technologies are integrated. For instance, quantum algorithms may completely alter and revolutionise machine learning. Faster data processing is made possible by quantum computing’s ability to do concurrent multi-state operations.
You can improve your data analysis and acquire deeper insights by using Quantum ML. Companies may get greater outcomes with these advancements than with traditional ML techniques.
There are no commercially accessible quantum ML models as of now. However, major IT firms have begun to invest in this technology, and the emergence of quantum machine learning systems is imminent.
Raheem noted that Pakistan has a long way to go to deploy ML and increase productivity in industries like healthcare and agriculture, which are in dire need of technology advancements. Raheem was speaking about the future of Pakistan.
“Although talent and skills exist, they have not yet been put to use in Pakistan to improve conditions. Pakistan should establish an AI strategy if it wants to compete in the international economy. The nation has not yet benefited from the fourth industrial revolution and is already trailing behind the rest of the world in the use of AI. It would be devastating for the country’s efforts to miss the AI wave, “Finally, he said.