Continued from Artificial Intelligence for Children – Part 1
In our story, we have reached a point where computers and internet are making changes to our lives in all possible way and we are waiting for Artificial Intelligence to make a grand entry. So, continuing from there.
Now, computers though ubiquitous and seemingly omnipotent, are dumb machines (no hard feelings, buddy). You tell them something to do and they do it.
Scientists always wanted these machines to act intelligent. Unfortunately, computers could not even differentiate between the image of a dog and of a cat (something, we, as humans, start doing as toddlers). Nor could the machines understand the context of what we said to them e.g. whether we are talking about Apple as a company or apple as a fruit, machines had no idea of what we wanted. Just imagine that you gave a lecture to a room full of machines on Apple being one of the most valuable corporations in the world and the machines murmuring among themselves that “do they need to find the stock price of Apple or provide the benefits of eating it”.
So, this has been the quest for a long time – how to make machines intelligent – how to make the computers mimic human behaviour and act sensible. While the research has been on since the days that the computers were born, it is a combination of three factors over the last two decades that started making this idea a reality –
- availability of huge amount of data
- tremendous increase in computing power of the machines and
- some very powerful algorithms coming out of research
Well, just in case you do not know, an ‘algorithm’ is a set of rules to be followed for solving a particular type of problem. Computers programs are implementation of various algorithms to solve different types of problems.
Let us look at each of these points a little closely. Data availability has been growing exponentially over the years. When you take a selfie, when you send an email, when you post a comment on facebook, you are creating new data.
These days machines also are ‘talking’ to each other (no, don’t think of Toy Story type of communication between machines) – machines are generating data through sensors that are monitoring performance of the machine on various parameters; or a modern day electric vehicle like Tesla and the self-driving cars that are being tested by various companies are generating a lot of data as they speed on the highways.
This information is being shared with other machines (mostly on ‘Cloud’ – these are computers stored in some remote locations very much on earth and not in the sky – and some of the largest Cloud service providers are none other than Amazon, Microsoft and Google) and this communication between machines is often referred to as IoT or Internet of Things (I know, there are so many terms floating around and it could get confusing).
Well, the point here is that there is a lot of data being generated today and this has been of tremendous help in making machines become intelligent.
The second factor is the way computing power has increased in our machines over the years. Our mobiles and smart watches are way smarter compared to the super computers that were crunching numbers for Apollo 11 in late 1960s (so, you can send a person to moon using your smartphone!). When you play a game on Xbox, have you wondered how well it renders images and how quickly to give a near real experience?
The third factor is the rise of the new algorithms. The Boss Babies in the world of algorithms can be categorized under ‘Machine Learning’ (a subset of Artificial Intelligence) and Deep Learning (which is a subset of Machine Learning). These algorithms started giving the machines the long-sought intelligence. We will discuss more of these in the next part of this article series.
Let us take a peek into how these intelligent machines are helping us today? We will take a few examples and see the impact that they are having on our daily lives.
- Watching movies on Netflix: Have you noticed the recommendations that Netflix makes for the series or movies that might be of interest to you? Are they the same recommendations that it is making to your aunt who lives in a different city? The recommendation systems for Netflix (and for a lot of digital content companies) are quite intelligent today. Before making a recommendation, it puts people of similar preferences, profiles, demographics into groups and assesses what content could a person in a group be interested in. This is different from the traditional recommendation systems that would suggest content based on the ‘most popular’ series without any personalization.
- Tagging friends on photos in Facebook and in Google Photo Albums: As I had mentioned, before the advent of AI, computers were not able to differentiate between the images of dogs and cats. But, today when you upload photos of your family after coming back from a recent vacation in your digital albums or on social media, the applications automatically tag your sister, your friends and your parents in the photos where they are present. You can later search through your albums to find only those where your best friend was present. How cool is that? This is all made possible because of AI.
- Typing a message or mail on your smartphone: Most of you have been using smart phones for at least 8-10 years. Just think that how has the experience of typing a text message on a phone changed. Earlier there was a QWERTY keyboard and Blackberry ruled the space. Now all phones offer a virtual keyboard on the screen itself.
It started when Apple started changing the size of on-screen keys (not visually but behind the scenes), based on the probability of what key a user will type after the previous key (e.g. after you typed ‘s’, keys ‘d’ and ‘f’ are less likely compared to ‘e’ and ‘a’). This unknowingly made life easy for users as the thumbs would normally span more than a single key when typing on the small screen.
The next change came in form of autocomplete – phones started offering the most likely words to the user in the middle of typing a word.
Taking it even further, using AI, phones are now offering predictive text. The phone has started showing three words just above your virtual keyboard—words that you are most likely to type next. For example, when you type “Best,” the phone offers “wishes”, “regards” and “of” as the likely options to choose from.
- Driving down to your cousin’s home who lives at the other end of the town: We are all using Google maps these days for navigation on the roads. While google tracks us on the road (and other vehicles too), in order to predict the time it would take us to reach our destination depending on the traffic conditions and road diversions, a lot of AI is running in the background.
There are a lot of other applications of AI these days – for example, the dynamic pricing of Uber cabs depending on the supply and demand and the time of the day, and, detection of fraud transactions in banks and identifying cancer cells in CT Scan reports and so on.
We will look at some of these industry use cases of AI in more details in the coming sessions and we will also try to peek into the approach that is taken to make the machines behave intelligently.
Well, it is possible that before you started going through this series of articles, you were of the impression that AI is all about amazing robots and machines with superhuman capabilities. The use of AI in the above real-world scenarios, might have been a dampener. For you, there is some good news – there are companies that are working on self-driving cars, humanoid robots and trying to make very interesting ‘Hollywood style’ machines.
Irrespective of the types of these applications, I think you must have realized that AI is already all around us and it is going to have a significant impact on our lives in the coming years.
Now that we have got an idea of what AI is and how it is being used in the industry today, let us look under the hood of how AI works. However, for that you will have to wait for the next article in this series. Till then, goodbye!