Machine learning: Teaching machines to learn from humans

• Author:
• Author name

  Quantumrun Foresight
• September 1, 2022

Insight summary

Machine learning (ML) trains software to evolve by analyzing vast amounts of data, leading to applications in numerous fields. It falls into two categories: supervised, where the computer learns from labeled data, and unsupervised, where it discovers patterns in data without specific outcomes. These technologies not only enhance efficiency in various sectors but also drive innovative solutions, such as in predictive oncology and network security.

Machine learning context

Machine learning has enabled some of the most powerful technologies available today by allowing software and digital tools to be trained and progress over time. ML focuses on designing and developing algorithms that allow computers to analyze large volumes of information (big data) and learn from it. The application of ML has increased due to the availability of big data and the continuously improving computational power of computing hardware. 

There are two prominent types of machine learning: supervised and unsupervised. In supervised learning, the computer is allocated a training data set, and its task is to learn a model that can be used to construct predictions on new data. This type of machine learning requires the information to be labeled, meaning there is a known output for each input. In unsupervised learning, the computer is assigned a data set but not advised about the outcome. The task of the computer is to discover a structure or pattern in the data. This latter methodology is where ML truly shines by allowing computers to analyze information over time and identify emerging behavior or potential action.

According to a 2021 survey from consultancy firm PwC, 86 percent of respondents indicated that their company heavily uses AI technologies. In fact, during the COVID-19 pandemic, AI/ML tools proved to be highly beneficial, and more than half of the PwC survey respondents plan to accelerate AI/ML adoption plans. 

Disruptive impact

Machine learning can quickly organize and categorize large databases over extended periods of time so that algorithms can continue to scan and analyze data at any hour of the day. Additionally, machine learning improves with exposure to information and repetitive tasks. For example, in predictive oncology, ML can scan thousands of patient databases, hundreds of tumor types, and over 20 types of cancer. A machine learning algorithm can then compare potential medications based on real-life results. This information enables researchers and oncologists to access a detailed reference of optimal treatments based on different conditions. 

Similarly, machine learning may have far-reaching applications. For instance, AI/ML may help monitor an organization’s network security. This technology may scan connected devices rapidly to identify and flag risks before exploiting an organization. 

ML may reduce costs and time for organizations by automating procedures and identifying waste in supply chains. According to a PwC survey, 75 percent of management teams now rely on AI to make data-driven business strategies. In addition, 75 percent of firms that embraced AI solutions said they were able to continue innovating and improving their products and services to fit customers’ needs. According to Nasdaq, the AI/ML market is expected to grow by USD $20 billion from 2021 to 2025.

Implications of machine learning

Wider implications of machine learning may include: 

• Accelerated drug and vaccine development, resulting in quicker availability of medical treatments.


• Automated diagnosis, patient care, and treatment management, enhancing patient outcomes and efficiency.


• Refined advertising strategies and highly tailored products and services, increasing consumer engagement and satisfaction.


• Inventory and supply chain automation, leading to improved efficiency and cost reduction.


• Automakers integrating AI/ML in self-driving cars, significantly reducing accident rates and improving road safety.


• Financial institutions applying ML for fraud detection and risk management, substantially lowering instances of financial fraud and enhancing customer security.


• Personalized learning experiences, resulting in improved educational outcomes and reduced learning gaps.


• Predictive analytics in stock management, leading to optimized inventory levels and reduced waste.


• Governments implementing machine learning for urban planning and public services, leading to more efficient and responsive city management.


• Renewable energy optimization, contributing to a reduction in carbon emissions and advancement towards sustainable energy goals.

Questions to consider

• What might be the potential risks in machine learning?


• How do you think this technology will further change how society or industry processes data?