If you are a coder, you probably think Machine Learning is the most important thing since sliced bread and AI is just hype. However, if you aren’t a coder, you may think they are the same thing. Well, let me explain…
Let us start with that simple question, “What is intelligence?”.
What Is Intelligence?
Human intelligence is how we humans, as the most intelligent animals on earth, manipulate our environment to achieve a goal. If society’s widespread adoption of AI gives us a moment to re-imagine everything, what does this mean for humanity? Can we build machines to help us achieve our goals? If so, then would artificial intelligence help us do this?
The European Union (EU) has written guidelines on the ethical use of AI. They have emphasised that AI (which includes ML) should be human-centric, have an ethical purpose and be technologically robust and trustworthy.
We see here that the EU’s ideal goal has built within it: ethics - human values, principles, and rights.
It acknowledges human subjectivity as a fundamental part of an AI system.
More recently, Stuart Russell, in his book, argues that human-compatible AI is a way to ensure AI compliments humans. One example Stuart Russell gives is that we ask AI to determine what is best for us.
This is not as easy to do as it is to say. It is a much more complex problem, and we cannot afford to focus solely on the machines.
AI is a more complete and compelling approach because humans are the focus, the teachers, the creatives, the judges and perhaps most important of all, the goal setters. AI asks us to understand how intelligent entities, humans and machines, will interact with an environment.
What Is Machine Learning?
ML is a fundamental and significant part of the digital revolution and has an almost century-long pedigree. Today, it’s popular and used by millions of scientists, engineers and mathematicians.
Businesses in the digital and ML arena are big, really big. We often hear the term Machine Learning used interchangeably with AI. This slip of the tongue misses the real deal of AI, and the next step beyond ML is doing real AI. ML will play a part in this, and the existing ecosystems are already rich with useful capabilities.
What Is Artificial Intelligence?
AI is the study of how humans use intelligence to achieve goals. Humans are nature’s most intelligent mammals. We, humans, are subjective and objective beings. We are also conscious, which makes us, along with other animals, unique in our approach to achieving goals. For us to compute, we need to be conscious.
Machines, just because they can compute, do not need to be conscious. Even if a digital machine simulates consciousness, the digital machine is not conscious. If our current AI is limited to a digital computing ML offering, however impressive, we are missing out on a lot more that AI can do for us. We, humans, are ethical, value and principle-driven subjective and objective intelligent animals. We are the intelligent entities on which AI is based and are far more than digital computers. Intelligent entities of the future will be humans and machines.
Where Machine Learning Sits Within Artificial Intelligence
ML is part of AI. We certainly do not want to lose this progress. However, the focus on machines (digital computers in particular) and data loses that key missing element, us humans. Machine Learning is about learning from data. The way we achieve our goals is not a sterile objective scientific analysis of data. Instead, we experience our goals and use that experience to work with others and machines.
And, if we want to teach others, build products or manipulate our environment, we need a lot more than data and digital computing. If an engineer is designing a car, then Machine Learning would be one of many options available to solve that problem. Professional engineers would define an objective and scientific problem that we have a reasonable chance that a machine could help us with. Machines are needed to help us with our science. We can’t, for example, analyse all the scientific data we have from our experiments!
We also see these Machine Learning techniques helping in our search of the internet. Again, super-human capability with machines helping us every day.
More Than Just The Sum Of Its Parts
Consider and reflect for a minute. How do you teach concepts like charisma, capture tacit knowledge, or, on a more practical level, engineer a robot or an intelligent product or entity?
Is rapport simply a mathematical model that can be solved on a digital computer? Unfortunately, our ML data-driven analysis has limitations, and we need to go back to AI.
Stuart Russell and Peter Norvig have written the leading textbook on Artificial Intelligence, and they describe AI as the universal subject. This doesn’t mean it can solve all problems, but it can help us achieve our goals. Machine Learning is about learning from data, and AI is about learning from experience.
AI asks us to think about humans and machines and understand where is the learning from experience. AI also looks at how our intelligent entities interact with an environment. Again, this understanding is far more involved than simply analysing data on a digital computer. The environment which AI asks us to understand could involve humans. We, humans, are much more involved than digital data!
If our intelligent entities interact in an environment, we must understand the nature of these interactions. To do this, we need to go way back to the base of natural sciences and data science too. We need to think about objects and how they interact with each other. Aristotle first formed the basis of this, and his original ideas formed the basis of what most universities now teach - natural sciences, mathematics, data science and the list goes on. It led to the scientific method and ontologies, and this has underpinned the industrial revolutions. The third industrial revolution embraces digital computing, and we are now far into the fourth industrial revolution that includes AI.
AI is about intelligent entities of which humans are a crucial part. So let’s take Machine Learning forward by putting the adult (us humans) back in the room!
Hopefully, we have raised more questions than answers meaning that you are learning from your experience, and you want to learn more about AI.
If so, we have a range of Artificial Intelligence (AI) training courses for you, and the good thing is that you don’t need to learn to code.
The BCS AI Foundation Pathway is also currently in development. You will be able to choose from 12 awards set within four key areas of AI:
Each of the 12 awards is a qualification in its own right, carrying between 3 and 5 credits.
If you take four awards (20 credits), one in each key area, you’ll get the BCS Artificial Intelligence (AI) Foundation Certificate.
And if you continue further along the pathway and take eight awards (36 credits) - you can achieve a Foundation Diploma In AI.