Machine Learning: Your smartphone is smarter than you think

Negotiating time zones is a science that is hard to master. Yes, it is simple mathematics, but that can be hard too, especially when you have been travelling for seemingly endless hours, subjected to questionable airline food and a neighbour who disturbed your rest by asking for movie recommendations.

I was visiting Nairobi to attend Google’s ‘Magic in the Machine’, a live demo event about the company’s giant strides in artificial intelligence and how machine learning will make our lives easier, smarter and more convenient.

It was a tightly packed trip that lasted 48 hours in a city that was very chilly. My hotel room had no windows, so there was no allowance for fresh air. I soon developed a chest infection, a wheezy cough and a burning fever.

I was vaguely aware that my return flight to Lagos was at the crack of dawn – at 7am. But I wasn’t sure: Was it 7am Nigerian time or Kenyan time? Kenya is two hours ahead of Nigeria. I was too preoccupied with getting medication for my fever to check and make sure. The night before my flight, I opened Google Now and saw a notification. The search service had parsed through my e-mail, presenting my flight information as a notification. It highlighted and reminded me of my flight time per my location and pointed out a detail I had completely missed: my flight terminal.

Read: Report yourself: the citizen v. corruption

I had not realised that Jomo Kenyatta Airport had more than one terminal (I was too used to Murtala Muhammed International) and would most likely have ended up at the wrong one had I not seen that notification.

Saved by Google Now

Google Now had delivered two important pieces of information, one that I was not sure of previously and one that I did not know I needed, at a time when my judgement was impaired by being sick.

How did the application do this? Artificial Intelligence via machine learning, Google’s preferred approach to developing the former.

With the Google Assistant, fiction has become real life.

“Artificial intelligence would be the ultimate version of Google. The ultimate search engine that would understand everything on the Web. It would understand exactly what you wanted, and it would give you the right thing. We’re nowhere near doing that now. However, we can get incrementally closer to that, and that is basically what we work on,” said Larry Page, cofounder of Google, in 2000.

Seventeen years later, the search engine has brought us all significantly closer to a smarter world in ways we might not even recognise.

Google maps. Photo: Andrew/Flickr

Have you ever typed in a search term and before you have finished, it is completed for you? Most likely, the answer is yes. When you enter a certain location repeatedly as a destination on Google Maps, have you ever seen a suggestion that you save it as ‘home’ or ‘work’, because the software is estimating that such a regular destination might be one of those two?Most likely, yes. When you shop online or search for certain items over a period of time, have you noticed how you start seeing advertisements that pitch those products to you? And if you use Google Now, you start to see news about such search terms in your feed.

What is Machine Learning?

Artificial Intelligence (AI) is the science of making things smart, as is illustrated by the above examples. Machine Learning, the tool that AI uses, is teaching a computer to recognise patterns, such as repeated Google Maps destinations or repeated searches for certain items.

In Nairobi, I had a sense of just how smart the future might be. The Google team, consisting of machine-learning expert and principal scientist at Google, Blaise Aguera y Arcas, and Laura Scott, a communications manager, delivered a live demonstration and a talk on the company’s latest advancements in machine learning. Two of the demonstrated applications were experimental: the AI Duet, a piano that responds to you, and the Quick Draw, a doodle recognition game that predicts what you are trying to draw. It was immediately apparent that the latter could be a hit in a children’s art class and a great way to introduce them to technology.

Google HQ. Photo: Ben Nuttall

The most striking advancement was the Google Assistant. (You can download this via the Play Store on Android or Apple Store for iPhone.) Think of every science fiction film you have ever watched where actors talked to robots and they talked back. They could tell the weather, take instruction to place calls, they might even recommend a cocktail pairing with your dinner.

With the Google Assistant, fiction has become real life.

This virtual companion – an app in your phone brought to life by your voice – can carry on a natural conversation. Scott had a ‘chat’ with the app about when her next flight was, what airline she was flying and what terminal she was supposed to be at.

Eventually, machine learning will do more than just making sure you need not fork out any extra money for a new flight ticket.

Read: Top 10 nominees for Innovation Prize for Africa announced

According to Martian Herald, an online directory on unusual facts, machine learning will play a huge role in our finances and health. Frankly, it could help preserve our lives. Machine-learning software will help protect consumers from fraud by spotting changes in spending and credit card use. Anything unusual and you might receive an alarm in your inbox. Special software is in development that can help detect heart attacks before they occur. There are intelligent devices that can help distinguish between life-saving and fake drugs. Bear in mind that according to the World Health Organisation, 10% to 25% of medicine in developing countries is counterfeit. In 2015, the National Agency for Food and Drug Administration and Control (NAFDAC) impounded N270 million worth of fake drugs. News reports of Nigerians dying due to expired or fake drugs are all too familiar.

University of Cape Town (UCT) scientists work in the Drug Discovery and Development Centre (H3-D) laboratory in Cape Town, South Africa, 30 August 2012. Photo: ANP

Software that can track changes in health records and so help doctors to diagnose patients more efficiently is also being developed. Consider that in many Nigerian hospitals, every second ailment is interpreted as malaria. This stubborn misdiagnosis results in many deaths that could have been avoided.

There are intelligent devices that can help distinguish between life-saving and fake drugs.

In addition, machine learning software is in the works that can help us prepare better for adverse weather conditions, such as the flooding that was recently experienced in parts of Nigeria. Sure, that is not a substitute for the availability of well-maintained drains or competent urban planning, but if you knew that a disturbing amount of rain was on its way, you could act in accordance with that information and get your belongings and loved ones out of harm’s way.

It is comforting to know that one company – Google –  has made it its business to think along the lines of making all our lives simpler, smarter, safer and more secure.

Machine Learning is definitely a science I will be studying.