C-3PO, R2, replicants and Terminators, what’s the link? Two words condensed into one: artificial intelligence (AI).
What is Artificial Intelligence?
True to its name, artificial intelligence is man-made intelligence; intelligence created in a lab that aims to mimic natural intelligence, which is what we and other animals possess. Like R2 and Threepio, we usually think of AI as a conscious robot who can feel, think and act like a human. While this may someday happen, the reality is that we’re still very far from it. That’s not to say that we’re not well on our way — we’re making strides every day that are taking us closer and closer to our dreams (or nightmares?) of machine consciousness.
Before we get there, though, let’s talk about AI as we have it today. Many people may not even realize it, but features like auto-suggestion and voice to text are what makes up the artificial intelligence we have today. Sure, we also have self-driving cars and facial recognition, but the AI of today is very singular — it does one thing very well, and only that one thing. It can ‘learn’ to do more, but it doesn’t have the capability to bring it all together. For example, a Tesla Model S may able to drive without human input, Siri may recognize human speech, and the iPhone X can distinguish between different faces, but they’re not all interchangeable.
As leading digital marketing agency Aumcore, mentioned: AI has had a pretty big impact on our society. Someday soon it may even surpass our own intelligence, but that’s all speculation and, until then, all we can do is use it to our advantage. Whether it’s voice technology, predictive analysis or a chatbot, it’s safe to say that AI is here to stay.
Underlying AI is the same basic principle: learning from use. For example, for facial recognition to work, such as the iPhone X you may be using to read this, the machine ‘learns’ your face by first scanning it and learning its features. Delving deeper, the magic happens with a dot projector that beams out more than 30,000 invisible infrared dots, and an infrared camera that captures your face. Combining both, they’re pushed through a neural network to create a mathematical model of your face that serves as a template. When you want to unlock the phone, your face is checked against the mathematical model captured earlier, and if it’s a match, boom, the phone is unlocked.
Sounds cool, right? The question now becomes, how did we get here? How did we go from conjecture to reality?
When Did Artificial Intelligence Come into Existence?
The term itself — artificial intelligence — was coined by John McCarthy in 1956. A computer scientist by trade, he and fellow scientists, Claude Shannon and Nathan Rochester, proposed that “every aspect of learning or any other feature of intelligence can be so precisely described that a machine can be made to simulate it.”
Was this trio the first to think of what artificial intelligence is? Nope, not by a long shot. The ancient Greeks had many stories of inanimate objects being endowed with life, such as Pygmalion’s statue (who he fell in love with —> ancient version of Her?), Jewish folklore includes golems, animated clay or mud anthropomorphic beings who are similarly endowed with life, 16th century alchemists sought to create a homunculus, a small human being, Mary Shelley famously wrote Frankenstein, a story about literal man-made ‘monster’ with thoughts, feelings, and emotions, and so on.
Fastforwarding a couple of centuries and going back to AI as it’s used today, we can consider Alan Turing’s Computing Machinery and Intelligence, which introduced us to the Turing test, as one of the first building blocks of artificial intelligence. His paper deserves its own post, but for now we’ll suffice by saying that in it, Turing questions whether machines can think, and then determines that if a machine could carry a conversation that was indistinguishable from a human conversation, then it can be said that it was ‘thinking.’
A lot happened before and after, such as Vannevar Bush’s essay, As We May Think, which was an expanded version of an earlier essay (Mechanization and the Record) in which he talks about a machine that uses lower level technology to achieve higher level organized knowledge, and the Ferranti Mark 1, the world’s first commercially available computer, which was used to write a checkers a chess program — something knows as Game AI, but we’ll condense it by saying that from then to the present day, AI had it’s good and bad years.
The good years, AKA the golden years, were marked by discovery and technological advancements, and the bad years, now known as AI winters, were caused by financial setbacks and demoralization. We can also divide the time between the 1950s and the present into three separate eras of AI:
- 1950s – 1970s: Neural Networks
- 1980s – 2010s: Machine Learning
- Present Day: Deep Learning
The Four Types of AI
From the three eras come the four types of AI we’ve encountered and will encounter.
Type 1 AI: Reactive Machines
Like the names states, Type 1 machines are purely reactive — give them an input and they produce an output. These machines, like Deep Blue, IBM’s supercomputer that beat grandmaster Garry Kasparov at chess, ignore everything except for what’s in front of them in the present. In Deep Blue’s case, it looks at all the pieces and makes a prediction about what its opponent will do, and then moves accordingly.
Type 2 AI: Limited Memory
Type 2 machines add the past into the equation that leads to action. For example, Tesla’s self-driving cars monitor other cars’ speeds and direction over time, and use environmental inputs such as lane markings and traffic lights to decide how to react in the many different scenarios it may encounter.
Type 3 AI: Theory of Mind
From now on, we’ll be discussing the future, as we haven’t arrived at Type 3 AI yet, much less Type 4. Type 3 machines, like ourselves, will evolve to be able to understand other people, creatures and objects, and use that understanding to direct their own behavior, hence the Theory of Mind name.
Type 4 AI: Self-Awareness
And now we arrive at self-awareness, the stage that the likes of Elon Musk and Stephen Hawking have warned us against time and time again. Why? Because with self-awareness comes consciousness and the ability to think. These machines will be able to know their own internal states, and will therefore be able to understand and predict others’ as well. The problem is, what will they do with that knowledge?
That’s All… For Now
Well, that’s it, that’s the gist of it — the basics of artificial intelligence. And yes, I know, I know, we ended it on a rather ambiguous question of what AI derived consciousness entails. If you want that question answered, stay tuned for the next AI themed post!