How to explain Machine Learning to your grandparents

Original article was published on Artificial Intelligence on Medium

How to explain Machine Learning to your grandparents

The main purpose of this article is to explain a complex topic which is Machine Leaning in the simplest way possible, so everybody can understand the concept even those who are not into technology. Let’s start with the basics reviewing basic concepts and where is the term Machine Learning coming from.

Machine Learning is one of the applications of Artificial Intelligence that creates systems that are able to automatically learn by themselves, based on their experiences, without being explicitly programmed [1]. It will do so by the study of algorithms and statistical models which enable a computer systems to make decisions and predict outcomes without being explicitly programmed [4]. We can understand Artificial Intelligence as a whole and Machine Learning as a subset of that whole, as shown in the image below.

The development of thinking machines, may sound amazing and scary at the same time, Artificial Intelligence (AI) brings out a lot of passion and many people have strong opinions. Some people feel that Artificial Intelligence is one of the greatest threats for the humanity. On the other hand, others believe is the path for a new world full of opportunities.

Either way, all we know is that AI is in its very beginning of what it can accomplish and will be shown in the following years. In this moment with Artificial Intelligence we have the ability to find patterns in sets of data, that decades ago would had been impossible to achieve. Scientists can now make better predictions in different fields, like the weather or like pharmaceutical and medical treatments. Businesses nowadays are using Artificial Intelligence to predict the buying behavior of their consumers, to have a better understanding of their requests.

Now that we have some idea of what is Artificial intelligence, Machine Learning and it’s relationship, let see where is the name Artificial Intelligence coming from reviewing the history.

Artificial Intelligence History

Back in 1955, Professor McCarthy coined the term “Artificial Intelligence” in a Dartmouth conference. This conference was to see if early computers could behave in ways that everyday people would identify as intelligent. Back them, computers were really big, but even with their enormous size, they had so much less processing power than modern smartphones. So they didn’t make much progress creating a fully intelligent artificial person, but they created a term that ignited everyone’s imagination. This opened a door to a new area of research for computer scientists. In the infographic below we can see the different developments in the AI field and how the leaded the progress until now.

· 1950 — Turing Test, Computer scientist Alan Turing proposes a test for machine intelligence. If a machine can trick humans into thinking it is human, then it has intelligence.

· 1955 — AI Born, Term “Artificial Intelligence” is coined by computer scientist, John McCarthy to describe “the science and engineering of making intelligent machines”.

· 1961 — Unimate, First industrial robot, Unimate, goes to work at GM replacing humans on the assembly line.

· 1964 — ELIZA, Pioneering chatbot developed by Joseph Weizenbaum at MIT holds conversations with humans.

· 1966 — SHAKEY, The ‘first electronic person’ from Stanford, Shakey is a general-purpose mobile robot reasons about its own actions.

· 1997 — Deep Blue, a chess-playing computer from IBM defeats world chess champion, Garry Kasparov.

· 1998 — KISMET, Cynthia Breazeal at MIT introduces KISmet, an emotionally intelligent robot insofar as it detects and responds to people’s feelings.

· 1999 — AIBO, Sony launches first consumer robot pet dog AiBO (AI robot) with skills and personality that develop over time.

· 2002 — ROOMBA, First mass produced autonomous robotic vacuum cleaner from iRobot learns to navigate and clean homes.

· 2011 — SIRI, Apple integrates Siri, an intelligent virtual assistant with a voice interface, into the iPhone 4S.

· 2011 — WATSON, IBM’s question answering computer Watson wins first place on popular $1M prize television quiz show Jeopardy.

· 2014 — EUGENE, Eugene Goostman, a chatbot passes the Turing Test with a third of judges believing Eugene is human.

· 2014 — ALEXA, Amazon launches Alexa, an intelligent virtual assistant with a voice interface that can complete shopping tasks.

· 2016 — TAY, Microsoft’s chatbot Tay goes rogue on social media making inflammatory and offensive racist comments.

· 2017 — ALPHAGO, Google’s A.I. AlphaGo beats world champion Ke Jie in the complex board game of Go, notable for its vast number (2170) of possible positions.

· 2019 — Google’s AlphaStar defeats professional Starcraft II players.

In the timeline above we have seen computers have been beating human players of different games and competitions, but no matter how good they could be, none of these computers understand the purpose of the game or even why were they playing, they are simply using their talent of following rules and matching patterns. So is important to be aware that human intelligence is different to computer intelligence, meanwhile humans can reflect and think out of the box, computers are unaware of the concept of a game they can win, and are not even aware they are good at it, they just can be so much better than any human in identifying and matching different patterns.

It’s important to notice, that the computers can shine and show their intelligence in the perfect environment, one that has a set of rules with a certain amount of possibilities, so it can use pattern matching to check its database for a possible answer. So in terms of applicability of AI in business, the organizations that can get the most benefit are the ones that work with a well-defined space with a set of rules. For example Google, which business is all about pattern matching, they match user’s questions with their massive database of possible answers. So if we have a lot of pattern matching in our organization and a set of rules and possibilities, then we can get benefits form Artificial Intelligence.

Types of Artificial Intelligence

In 1980, the philosopher John Searle presented the “Chinese room argument”, which he pointed out that we can think of Artificial Intelligence in two different ways, “Strong AI” and “Weak AI” [3].

Strong AI, is when we have a machine that displays all the behavior that we would expect from a full artificial person. Machines that would be able to perform a multitude of tasks of interest to a similar level as a human and is indistinguishable from a person in the Turing test [4]. This is usually what we see in science fiction movies, these artificial beings that have emotions, sense of humor and sense of purpose, this is also called “General AI”.

Weak AI on the other hand, which is often called “Narrow AI”, here we have a machine that is confined to a very narrow task, maybe to try to process a natural language into text or trying to organize all the pictures on our computer. So a weak AI program is not trying to talk with us about the world, it’s just trying to assist us with a very narrow task. In other words, Artificial Narrow Intelligence performs a limited function or set of functions extremely well, often better than a human, for example image recognition or autonomously driving a vehicle [4].

Here is where comes to place AI assistants like Apple’s Siri, which are a good example of weak AI. We can talk to them and even ask them questions, and they will convert our language into something that computers can recognize. They will also do pattern matching to connect our request with their database, which is not much different from when we type something to Google, the difference is that it seems to behave much more like a human. This personal assistants will search the web, our smartphone, our contacts or calendar, and perform operations in return, like calling a friend, scheduling a meeting and so on.