top of page

Unpacking Artificial Narrow Intelligence: Dive into the key subfields—Machine Learning, Deep Learning, and Neural Networks



Artificial Narrow Intelligence (ANI), also known as ‘weak AI’, has been successfully realised and has been in existence since the 1950s. This is the only form of AI that has been achieved thus far.


ANI is the broad concept that encompasses numerous subfields such as, machine learning (ML), deep learning (DL), and neural networks. ML learns patterns from data, while DL leverages deep neural networks for intricate pattern recognition.


Machine Learning

Machine learning is a subset of AI that focuses on developing algorithms and models that enable computers to learn from data and make predictions or decisions without being explicitly programmed to do so. It involves techniques such as supervised learning, unsupervised learning, and reinforcement learning. There are numerous models that exist within machine learning: Computer vision, natural language processing, speech recognition, predictive analytics, and robotics, to name a few, that are commonly known. However, scaling a machine learning model on larger data sets often compromises the accuracy. Another major limitation of ML is that a human needs to manually figure out the relevant features of the data, based on high level knowledge of the data, and feed that to the machine for training.


Deep Learning

Deep Learning is a subfield of machine learning, and a technique that involves neural networks with three or more layers (hence the term “deep”). Each layer passes on a more abstract representation of the original data to the next layer, with the final layer providing a more human-like output. DL can manage complex tasks and larger datasets far more efficiently; automatically extracting relevant features, discovering intricate patterns, and eliminating manual feature engineering, becoming further removed from the human input. DL emerged as a result of the limitations of ML.


From the advancements of Deep learning, emerged Generative AI, a more sophisticated subset that harnesses input data and employs pre-trained transformative algorithms (like GPTs) to produce new content.


Neural Networks

Neural Networks are computational models inspired by the structure and function of the human brain. They consist of interconnected nodes (or neurons) organised into layers. Each neuron performs simple mathematical algorithms based on its inputs and passes the result to subsequent layers. Neural networks learn by adjusting the strengths of connections (weights) between neurons during the training process.


The terms AI and machine learning are frequently employed interchangeably, or as blanket terms, when referring to any of the subsets. Additionally, AI is used as a broad term when discussing any complex computer automations. This flippant use of the term has led to heavy criticism of the term by field specialists.


‘True’ Artificial Intelligence is the concept of a machine having the ability to mimic human intelligence and cognitive functions, beyond problem-solving, reasoning, and learning. ‘True’ AI is when a machine becomes self-aware, self-taught and can mimic the complexities of human emotion. This is, however, still hypothetical. Current AI systems primarily exhibit narrow or ‘specialized’ intelligence, focusing on specific tasks or domains that are prescribed from human input.


Generative AI

Generative AI is a broad sub-set of Deep Learning that has shown the most potential in re-shaping industries. Unlike ‘traditional’ AI that is used to analyse data sets and make predictions or find solutions, Generative AI is trained on massive sets of data to learn underlying patterns, analyses those data sets, and generate new content based on its pre-trained knowledge. This new content includes text, audio, code, images, simulations, and videos.


Recent advancements in Generative AI from OpenAI led to the development of the GPT, Generative Pre-trained Transformer. GPT models, particularly the Transformer architecture that they use, represent a significant breakthrough in AI research.


ChatGPT and DALL-E, have drastically changed our approach to content creation. The value of these models lies in the speed and scale at which they operate.


For the most part, AI has played a periphery role in our lives, assisting us with tasks we may not have even been aware utilised machine learning. However, with the recent advancements made in Generative AI by the likes of OpenAI, the general public has become far privier to the role that AI is, can and will play in our day-to-day lives. The adoption of AI has more than doubled since 2017 and the proportion of organisations using AI is between 50 and 60%.


Armed with new capabilities and prospects of previously ‘fictional’ advancements in technology, industries are racing to integrate the latest AI systems into their business, and with increased social awareness of the capabilities of AI, customers have raised expectations for service delivery as a result.


In practical terms, AI technologies are being increasingly utilised across various industries and applications, including:


  • Natural language processing and chatbots for customer service and virtual assistants.

  • Image and speech recognition for medical diagnosis, autonomous vehicles, and security surveillance.

  • Predictive analytics for personalised recommendations, fraud detection, and financial forecasting.

  • Robotics, automation and virtual reality for manufacturing, logistics, and household tasks.


As AI continues its rapid advancement, its profound impact on society, the economy, and daily life is anticipated to increase significantly, reshaping our methods of work, communication, and interaction with technology.


Similarly, Enterprise Resource Planning (ERP) systems have experienced exponential growth, driven by advancements in technology and computational power, alongside societal demands for quicker, more optimised, and accessible work management solutions. Consequently, the evolution of ERP systems has naturally gravitated towards integrating AI over the past 15 years.


The debate is no longer about whether to incorporate AI into ERP solutions, as this has already been realised and implemented by numerous leading software companies. Instead, the focus shifts to how ERP systems can fully harness AI’s potential to not only benefit enterprises but also serve the wider public, both now and in the near future.

bottom of page