Artificial Intelligence (AI) has been a hot topic over the last five years. November 2022 witnessed the breakthrough of AI technologies like ChatGPT, and future projections for this field are optimistic. Morgan Stanley’s Global Investment Office sees AI as one of the most important investment themes of the next several decades and estimates a rapid growth to a $3 trillion industry in the coming years.1
AI is involved in decision-making and predictions through deep and precise analysis of data and adds more natural and efficient interfaces (such as conversational platforms) to existing products and technologies.2 AI is not automating the learning that humans do. Humans learn in a different way and we want them to continue to do so. Machine learning engineers need them to. AI automates the prediction of information. Machines are good at repetitive tasks – better than humans. So machine learning engineers try to make the world more efficient by getting them to repeat predictions of information. Machine learning (not AI) works by learning from examples of predictions in a training data set.
AI can be used in various sectors, including creating better diagnoses and treatment plans in the healthcare sector fraud detection and as investment advisors in the finance industry, and assisting with national security.3
However, AI is a complicated concept to understand, especially for non-professionals. A tricky thing about understanding AI is the massive gap between our intuitive knowledge of circuits and software and the ability for machines to “think”.4 In this blog, we’ll try to simplify the concept of AI so that anyone can understand it. We’ll explain what it is, how it learns, its relationship with machine learning, and the essence of building an AI model.
More than just a robot
Often, when we think of AI, images of robots and mechanical entities spring to mind. ChatGPT explains AI as a robot friend that needs to be taught what is right and wrong so it doesn’t make mistakes.
Even though the statement has merit, AI transcends the realm of physical form. A useful analogy is to consider it a guide on how to mimic being a human. By following instructions and learning from examples, it directs a computer in mimicking human thoughts and actions. Another way to understand AI is as a storybook or album stored in a computer’s memory about the past: it represents the vast knowledge about the world and helps the computer interpret and interact with references to its past.
In some sense, AI also resembles electricity as a magical, elusive, and emotive force that few understand. The prevalence of AI in recent literature is parallel to how electricity is introduced into common parlance at the turn of the last century.
Learning from examples and mistakes
Similar to humans, AI learns from examples and mistakes. Machine learning is a process when the machine’s method of doing something is refined by working through examples of what it should do, hence correcting its mistakes. This leads to a more precise model that mirrors the task it’s learning about, allowing them to make predictions on their own, which embodies the essence of AI.
However, there are still differences between how humans and machines learn. Unlike humans, machines cannot forget things unless they’re deleted by humans. However, it can be programmed to determine what’s important and what’s not. This categorization of importance is objective and should ideally be justified through human consensus. Moreover, AI might not connect examples in the same way as humans do, highlighting a difference in how we learn and understand the world.
AI and Machine learning
Machine learning makes models that can always be used for AI to answer questions or complete tasks, but AI doesn’t necessarily involve machine learning. One example is the earliest forms of AI from 1965, known as expert systems. They are based on human crafted rules that follow a simple “if X then Y” logic. These AI systems predate our understanding of even the smallest neural networks (a computer system designed to mimic the way human brains learn and make decisions) and mark the initial steps towards the complex AI technology we see today.
A blend of science and art
Artificial intelligence is meticulously crafted through human ingenuity. It involves a deep appreciation for the intricate layers of abstraction necessary to align countless 1s and 0s (binary codes of computer) perfectly. These layers upon layers of building blocks shaped the complexity and possibly fear of the technology – but when it comes down to it, it’s just 1s and 0s.
However, building AI is also a process requiring deep respect and careful observation of the world we seek to emulate. It’s a practice grounded in responsibility, authenticity, transparency, and accessibility, all while being fuelled by a relentless drive for creativity. Ensuring AI reflects our world with integrity and innovation has always been a fundamental objective at Cam AI.
I am a first year undergraduate student from Cambridge University studying Psychological and Behavioural Sciences. I am currently volunteering as an assistant at Cam AI, reading papers about AI and mental health, engaging in outreach activities and writing blogs. I am curious about the ways in which AI can enhance the mental health services provided to humans, and am very excited to be part of this team.