With tech companies investing billions into developing smarter and faster algorithms and some incredibly talented people working on how to create artificial intelligence, there are some really exciting things happening in a field that has the potential to disrupt everything we know about being human.
If you’re into being on the cutting edge of technology, business, philosophy, science, medicine.. basically anything that’s important in the world, then this is the time to get involved with artificial intelligence and all of its sub-disciplines and related fields – like machine learning, deep learning, robotics, and of course, the super on-trend field of data science.
What is artificial intelligence?
There is no universally agreed upon definition for the term ‘artificial intelligence.’ Instead we’ve ended up with an already complex topic that can mean different things to different people.
Hollywood is partly to blame for this confusion (I’m looking at you Westworld) with the production of super slick movies and TV shows that feature friendly humanoid servants or tyrannical robots intent on taking over the world and ridding it of us pesky humans.
At the same time, the field itself is relatively young and moving at a ridiculously fast pace so it can be hard to keep up, even if you’re following the latest happenings in the world of AI.
Because of this, it’s easy for us to think that all AI IRL is indeed fictional.
In fact, a lot of the time we don’t even realize that some of the systems we’re using multiple times a day are, in fact, AI.
From Alexa, Siri, and Google Assistant, to Facebook, Netflix, and email spam filters, to your car, cellphone or the last plane you flew in – these things (and so, so much more) all now use artificial intelligence in one way or another.
John McCarthy, the person who coined the term ‘Artificial Intelligence’ commented that “as soon as it works, no one calls it AI anymore“, which is probably true. Since these services are all now mainstream, talking about AI can feel like some crazy future prediction when in reality it’s all around helping us do things every single day.
AI can now carry out simple tasks like creating a customized playlist (just for you!) on Spotify or trolling Twitter, to seriously complex ones like discovering new uses for existing drugs or searching for extraterrestrial life.
Some people consider AI to be artificial life-forms that have super intelligence and are smarter than humans. Although in reality, we’re not actually at this point of artificial intelligence – yet.
Other people classify much of the commonly used data processing technology that’s currently out there as fulfilling the requirement to be considered a form of AI.
A pretty widely agreed upon definition of artificial intelligence is that it represents the use of a computer to model or replicate ‘intelligent’ behavior (such as the kind we might exhibit as humans), using algorithms to learn and perform this intelligent behavior with minimal intervention from us.
One #dadjoke definition is that AI should be thought of as “cool things that computers can’t do”, but this definition is pretty ambiguous.
Another way of defining artificial intelligence might be something that can act autonomously and adapt to its environment. So a system that can perform tasks in a complex environment without receiving constant guidance from a user while learning and improving from experience could be considered to be artificially intelligent. Like an algorithm that learns how to accurately identify pictures of cats by looking at millions of photos of cats.
The most basic break down of the term artificial intelligence would probably be something like:
“intelligence that is made or produced by humans rather than occurring naturally.”
But of course, this definition then opens up a whole different can of worms – for example, what exactly is intelligence and how should it be defined?
Moreover, how can you determine whether something is truly intelligent (i.e can do things on its own) or is simply imitating or copying intelligence?
Even in humans, the term intelligence isn’t something that researchers have been able to satisfactorily quantify. There is no singular metric or dimension that denotes higher or lower intelligence and unlike something quantifiable like weather in different cities, you can’t just compare one person’s intelligence with another and definitively say that person A has more intelligence than person B. Scientists have attempted this with things like IQ (intelligence quotient) scores, but these results are limited in what can actually be done with them.
If it’s hard to quantify intelligence in humans, imagine how hard it is to do for machines.
Would you say that the algorithm that beat the 18 time world Go champion is more intelligent than algorithms used for self-driving cars? It wouldn’t really make sense to compare the two and it would be hard to pin down in exactly what way one algorithm was or wasn’t more intelligent than the other.
This is further complicated by the fact that difficult tasks like complex math problems are incredibly easy for a computer, whereas what seems super-easy to us, like walking around, is incredibly hard for a computer to do without continual input.
Plus, there are a whole range of things about AI that even those working on the cutting-edge of research still don’t fully understand.
Fact: just because your iPhone can do something cool doesn’t mean it’s truly intelligent yet – sorry Siri.
Regardless, artificial intelligence is set to be a game changer and although scientists and researchers might have a hard time agreeing on a definition, the artificial intelligence revolution is underway.
Artificial intelligence: the basics
Scientists have outlined three different levels of artificial intelligence which are generally taken to represent the level of intelligence of an AI system:
Artificial Narrow Intelligence (ANI): aka ‘Weak AI’, artificial narrow intelligence refers to AI that specializes in one task, like playing chess or answering your search query. It’s unable to do anything beyond the singular task it was designed to do and lacks the ability to move laterally between tasks. Everything that we know of as AI today is ANI and, in most cases, tasks that have been outsourced to ANI systems tend to outperform humans in both efficiency and endurance.
Artificial General Intelligence (AGI): also goes by the names ‘Strong AI’ or ‘Human-Level AI’, artificial general intelligence is a computer that can do anything intellectually that a human can. Based on Professor Linda Gottfredson’s description of intelligence, AGI would be able to “reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience.” Basically, it would be as smart as we are. Scientists are hard at work on creating AGI – they haven’t cracked it yet but they promise that it’s coming.
Artificial Superintelligence (ASI): when some of the world’s biggest brains throw around phrases like “human extinction” and “worst thing to happen to humanity in history” they are probably referring to artificial superintelligence. Nick Bostrom, a leading AI thinker defines superintelligence as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom, and social skills.” One of the reasons people are alarmed at the prospect of superintelligence is that machines could go from just a little bit smarter than humans to trillions of times smarter in a matter of days or even hours. And because they would be so much smarter than we are, we’d have no way of understanding why or how they did what they did. There’s debate about how quickly (if at all) scientists will be able to create ASI but the prospect of it is something that keeps some of the smartest people in the world up at night.
Why artificial intelligence matters
As one NYTimes article so succinctly put it, AI “can probably do less right now than you think. But it will eventually do more than you probably think, in more places than you probably think, and will probably evolve faster than powerful technologies have in the past.”
Although AI most likely isn’t yet as powerful as all the hype around it would imply, it is already causing substantial changes to the way we work and make our money.
Automation has been and continues to be the first harbinger of this with actual robots already replacing dangerous and repetitive work, while smart tech means you don’t even have to make your own coffee anymore.
The scale of job displacement from automation is going to increase rapidly as the technologies in AI and its surrounding fields get better and better. Some estimate that from 75 million to 375 million people might need to switch jobs and learn new skills as a result of these technological transitions. To put that into context, approximately 375 million people speak English as their first language. That’s a lot of people.
Ultimately, artificial intelligence matters because one way or another it is going to affect you. AI means change and this change will bring with it uncertainty for many.
But here’s the thing about change.
Change also almost always brings significant opportunities – at least for people that are looking for them. And these changes? Well, these might just be the biggest changes humanity’s ever seen. In terms of scope, disruption, scale, speed… everything.
This really is a big deal.
You know the internet? This is bigger.
You know cell phones and smartphones? This is bigger.
You know social media? Bigger.
Imagine having a time machine and being able to go back in time. You’d know everything you know now but it hasn’t happened yet so you can bet on the Red Sox winning the 2004 World Series and invest in Apple and Microsoft in the early 90s to become one of the richest people in the world.
There are some things out there that don’t require a time machine to give you the insider tip though and the growth of artificially intelligent machines is almost certainly going to be one of them.
AI in context: how it all fits together
When it comes to AI, there are many different terms and buzzwords thrown about, so let’s take a look at what’s what.
The majority of topics covered on Elle Knows Machines fall either within the remit of computer science (making computers do things), mathematics (the language that describes all of the universe), or data (specifically, how to make machines utilize large amounts of it and then do useful things with it).
This site mainly focuses on the top-level categories of artificial intelligence and data science and in many ways these can be thought of as siblings – different but with a shared lineage and significant cross over.
In the case of AI, we’re really talking about a relatively broad term that encompasses a number of different sub-fields while being tangentially related to others, either directly or indirectly.
This is a visual conceptualization of how the different areas fit together:
If this is how the whole thing fits together, the next logical step is to gain a clear understanding of what each topic is and what it means as a part of the whole.
One of the terms that you’ve probably heard the most about in relation to AI is machine learning.
Machine Learning (also often referred to as ML) is a sub-field of artificial intelligence and refers to the area or discipline concerned with developing algorithms that make computers ‘learn’ a task or function without being explicitly programmed to do so.
The overall goal of these algorithms is to improve their performance in a given task as they receive and digest more data over time, ultimately helping AI systems to become more adaptive to complex environments and a wider range of known and unknown scenarios.
Deep Learning is a sub-field of machine learning and consists of specific machine learning techniques known as neural nets – basically algorithms inspired and influenced by the structure and function of the brain.
The ‘deep’ in deep learning refers to two things: the complexity of the mathematical model used to run these algorithms and the high levels of computing power now available that has allowed researchers to actually use such complex math. These two things combined are delivering results that appear to be significantly different from previous attempts.
Sitting alongside these areas of AI (not shown in the diagram above), are the fields which bridge the gap between the digital world and the physical, allowing intelligent machines to sense and interact with stimuli in the real world environment around them.
If the AI is the brain, then these areas are effectively the eyes, ears, arms, and/or legs, utilizing sensors to sense things from the environment and actuators to react to information and allow physical action.
The field of robotics is probably the best known of these and involves the building and programming of machines so they can do cool stuff in the physical world, utilizing a combination of artificial intelligence and machine learning, alongside sensors and actuators.
Robots are basically containers for AI and should probably be thought of as the physical manifestation of artificial intelligence (narrow or otherwise) that allow movement around and/or interaction with the physical world.
Sometimes robots look like humans and sometimes they don’t, but getting a robot to do things that humans can (in a natural way) is one of the holy grail tasks of this branch of study and includes many sub-areas and related disciplines like computer vision, natural language processing, and cognitive modeling.
There are some remarkable developments happening in the field of robotics right now, some of which you can see in this video:
Natural language processing
Although you may not know it, odds are you’ve already come across a well established form of natural language processing (NLP) through using a system like Alexa or Siri, or Google Assistant.
When you ask any of these systems a question, the software is using NLP to analyze and synthesize natural (aka human) language and speech.
This is a really important and exciting element of AI because it’s helping computers communicate with humans in their own language, whether that’s English, Bengali, or Mandarin.
Computer vision is another area of artificial intelligence that’s currently growing at an incredible speed.
As you’ve probably already figured, computer vision is about getting computers to be able to ‘see’ on par with human vision.
What began as a summer project assigned to a grad student, the challenge of true computer vision, has turned out to be incredibly hard to crack.
Still, researchers have been making some pretty big developments in the field. A pretty swish example of computer vision can be found on the iPhone X which lets you unlock your phone with your face or get emojis to mimic your expressions with Animoji.
But what about data science?
When it comes to making machines intelligent, data is a big deal and while large amounts of it are ultimately the fuel of data-hungry AI, the actual field of data science (collecting large amounts of facts and statistics and then obtaining actionable knowledge and insight from them) needs to be considered separately – find out how here.
The history of artificial intelligence
Here’s a brief rundown of the history of artificial intelligence so far, covering the major milestones in the field as it’s developed:
1956 – The foundations for artificial intelligence as a discipline were laid
What we think of as artificial intelligence began back in 1956 when Dartmouth professor John McCarthy invited a small group of scientists to get together over the summer and talk about ways to make machines do cool things. “We think that a significant advance can be made,” McCarthy wrote, “if a carefully selected group of scientists work on it together for a summer.” This workshop laid the foundation for researchers to form an academic field for people who dreamed about intelligent machines. In the beginning, work focused on solving abstract math and logic problems but it quickly became apparent that computers were able to perform more human-style tasks.
1962 – A computer was programmed to play checkers
One of the pioneers of AI was Arthur Samuel who created computer programs that learned to play checkers. He also coined the term ‘machine learning’. In 1962, Samuel’s checkers-playing program scored a win against a master checkers player.
1965 – The chatbot Eliza was created
Joseph Weizenbaum created Eliza, an early natural language processing computer program (aka chatbot). Eliza posed as a psychotherapist by paraphrasing the questions asked of it. Initially created as a way of showing the superficiality of communication between man and machine, Weizenbaum was surprised to discover how many people attributed human-like feelings to Eliza.
1975 – The first system to automate decision-making and problem-solving processes was created
A program called Dendral replicated the way chemists identified unknown organic molecules through interpreted mass-spectrometry data. Dendral was the first system to automate decision-making and problem-solving processes and also produced the first computer-generated discoveries to be published in a refereed journal.
1987 – The first self-driving car was debuted in Germany
Scientist Ernst Dickmanns kitted out a Mercedes van with two cameras and some computers which then drove itself along a German highway at 90 kilometers per hour (55 mph).
1997 – A computer beat world chess champion Garry Kasparov
IBM’s computer Deep Blue defeated world chess champion Garry Kasparov. Kasparov demanded a rematch but IBM declined.
2004 – The Pentagon launched a robot car race challenge
The Pentagon launched its first Darpa Grand Challenge, a race for robot cars in the Mojave Desert that kick-started the current self-driving car craze.
2012 – Deep learning renaissance
Advances in deep learning make speech and image recognition significantly more accurate, prompting a renaissance in the field fueled by new corporate interest.
2016 – Computer beat world champion Go player
AlphaGo defeated the world champion player of the board game Go, doing “a very human thing even better than a human.”
2017 – Deep learning system beat Ms. Pac-Man on Atari 2600
Using deep learning, AI company Maluuba created a system that learned how to reach Ms. Pac-Man’s maximum point value of 999,900.
Looking to learn more about artificial intelligence? Here are some guides to get you started: