How machine learning is already reshaping the world?

The term “artificial intelligence”, otherwise referred to as AI, can conjure up images straight out of science fiction, where computers fall in love and robots lead rebellions against humans. It can all seem extremely far off, if not completely unlikely, however, we might do well to think twice, as we are already living in a world where artificial intelligence is all around us.

It oftens presents itself in the shape of machine learning, a subset of AI, which, as Senior Data Scientist Rikus Combrinck explains, “is a set of mathematical techniques for learning patterns from large amounts of data in order to classify or predict things.” Machine learning is at a point now where it can handle unstructured data, of which the world currently abounds. In 2017, humans collectively produced more data in one year that in the previous 5000 years of humanity. By crunching large amounts of historical data, machine learning can help us to classify, predict or estimate, find similars, and create compound systems and generative models. 

But what does that actually mean? Practically speaking, machine learning can do everything from make diagnoses based on medical imaging, to detect fraudulent behaviour, predict who is likely to win a presidential campaign, make friend suggestions on social networks, identify and segment marketing target groups, mimic speech patterns and transform text into audio, drive cars, compose music, and even read minds. 

Needless to say the possibilities are staggering, and several multibillion dollar tech companies are already putting machine learning to notable use. For example, Facebook, the digital space where many of us spend a significant portion of our days, is also a platform where, unbeknownst even to ourselves at times, we produce copious amounts of data, ripe for the harvesting. The social media titan quickly understood the value of that data and currently uses machine learning in a variety of ways. For example, its automatic photo tagging uses state-of-the-art face recognition that already outperforms humans. Its Chatbots use Natural Language Processing, which turns natural language into structured data. The term “machine learning” couldn’t be more appropriate here, because the bots, who can not only parse conversational language, are also constantly learning from it and “getting "smarter" with each interaction”. Other examples of how tech companies are putting machine learning to use include Google, which uses the technology for automatic language translation, and Amazon, which uses it to make buying recommendations for its customers. Both tech giants also use it for speech recognition.

Because of the broad implications of machine learning, it would seem businesses are only limited by their own imaginations when it comes to the potential benefits that could be derived from it. South African based company OLSPS Analytics explains that “predictive analytics and machine learning helps businesses to step from (simply describing and diagnosing a particular phenomenon) to the more advanced levels (of analysis): predictive and, eventually, prescriptive.” In other words, machine learning can help us not just understand what happened and why, but also what will happen and what we should do about it.

As for what machine learning can do for the average Joe Soap, much of how it’s already being put to use is in aid of streamlining our experiences as consumers of goods and media, and as such, it tends to go unnoticed. If you’ve ever used Google Translate, clicked on an interesting series Netflix recommended, or used your fingerprint to unlock your phone, then you’ve already used machine learning in your day to day life.

There’s also a handful of quirky apps to play around with like Hot Dog Not Hot Dog (which very helpfully informs you whether what you’re pointing your phone’s camera at is or is not a hot dog) and Prisma (a photo-editing app which transforms your photos and videos into works of art using the styles of famous artists), but the most significant scope and potential of machine learning’s applications for the individual has yet to fully develop. There are glimmers of just how powerful it could be as a diagnostic tool in hospitals and as an everyday aid for people with physical and intellectual disabilities. Youtube is already using speech-to-text software to automatically caption speech and indicate applause, laughter, and music for the hard of hearing. Facebook has a feature that describes the image content of photos to blind people. There’s even a Chatbot that can help counsel people with depression.

For the most part, however, it seems as though machine learning is something that happens to people as opposed to for people. Our data is being collected and processed for the benefit of election campaigns, large corporations, and marketing agencies looking to increase profits. Here’s where the potential risks associated with machine learning start to crop up. Now that our data is a prized commodity, where does that leave privacy? If a computer can swiftly analyse, sort and classify individuals and identify, say, their sexual orientation in the process, just imagine if that information was used by a homophobic administration bent on “weeding out undesirables”, rather than by marketers who simply want to know how better to capitalise on our tastes.

When ghost-accounts on social media can share and promote misinformation, and uprisings can be identified and crushed before they’ve even begun, where does that leave politics? Proof has arisen about the extent to which Russian Twitter bots played a role in swaying the US elections that made Donald Trump president. "Any such activity represents a challenge to democratic societies everywhere," said Twitter. Trump’s rise to power has ushered in the era of Fake News that only looks set to worsen. With voice mimicry and video editing software that can reproduce speech patterns and facial movements, we are on the verge of an age where speeches can be remastered and locker room audio recordings can be fobbed off as fabricated.

Only time will tell if we have the capacity as a species to regulate our use of the truly spectacular technologies we’re developing at an ever increasing rate...