In the digital age, we are exposed to the early stages of artificial intelligence on a daily basis. Spearheaded by Google, machine learning is being used by companies to predict user behaviour and offer utility like never before. Entrepreneurs can take advantage of this burgeoning tech to build better businesses – but how far is too far?
The world of predictive search is here, and it is growing. From Netflix recommendations based on previous watches to Google search results based on your profile, algorithms are in place to sort and filter out information in a similar way to how our brains work.
On a basic level, algorithms are created to facilitate machine learning. They are strings of data fed to a computer to help it make decisions. But as well as helping you predict what to watch next on Netflix, they could also be used more controversially. For instance, your search preferences could be used to indicate if you will commit a crime in the future. In this way, the more advanced algorithms become, the more intrusive they are.
Businesses employing algorithms
Businesses are already using algorithms to boost their own profit – but entrepreneurs should be wary of the impact machine learning and personalised search is having on the user and the wider world.
When you visit an online store, you share details with the retailers you buy from. They generally learn your email address and know what you’ve bought. Target, the US retailer, uses this information as a ‘bucket’ of knowledge on their consumers.
Using this data, statisticians ran tests to see trends emerging. For instance, they noticed that in the first 20 weeks of pregnancy, women loaded up on calcium, zinc and magnesium. By using this data, they could predict when customers were pregnant and target their coupons and marketing towards each stage.
This led to a PR issue when the company predicted a teenager’s pregnancy before her own father knew about it. They sent her coupons for baby clothing and cribs, spooking them about just how much the company knew.
Algorithms in daily life
Law enforcement agencies use prediction to stop crime before it happens in the fictional story Minority Report, authored by Philip K. Dick. But the realm of sci-fi is expanding into modern day reality….
In 2014, the LAPD fed 13 million crime incidents to the University of California. From this data, an algorithm was produced that predicts areas where crime is likely to occur on a ‘mission map’. The trial of this algorithm saw a 36 per cent drop in crime rate.
However, reliance on this type of data can cause issues. An algorithm used by US parole boards can forecast the likelihood of a person committing a violent crime to help decide who to release and how to decide on an appropriate prison sentence. This system has a 75 per cent accuracy rate, which means it is wrong one in every four times.
How Google is pushing the boundaries of prediction
Google itself has openly stated it is a machine learning company. It uses algorithms to help your website rank better, giving birth to the world of Search Engine Optimisation. However, even the most talented marketers don’t really know exactly what goes into Google’s algorithm. Someday in the near future as it becomes more and more advanced, Google themselves won’t know. Search engines will be able to decide where a website ranks with no human input.
But search rankings are far from the only application that algorithms affect. Google’s AI systems are present in their advertising systems and their self-driving cars. If you have a Google account, Google can pull in data to ‘help’ inform search.
Typing in ‘my purchases’ to Google brings up Gmail receipts in your search box. Adverts are served to you based on your browsing history and preferences. Your calendar is pulled into results too. The engine knows if you’re visiting somewhere at a certain time, so if you search for restaurants in that area it will remind you of your trip there. But, as a post on digital marketing agency Mediaworks' website indicates, personalised results can become intrusive. A recent update seems to have trialled displaying a customer’s purchases alongside more generic searches. For example, searching for ‘capsule’ in hope of finding the CRM system of the same name may return your recent coffee capsule purchases.
However, personal results can be beneficial. In a world where habits shape 45 per cent of the choices we make, behavioural research and predictive analytics are gold dust for businesses. For entrepreneurs thinking about setting up a company, the world of machine learning and algorithm-based AI allows you to tap into vast knowledge. Your own data from purchases and analytics can help you make decisions that will boost your customer’s experience and increase the likelihood they will buy.
Google’s deep-learning future
As well as algorithm-based search, Google has an ongoing project that sees teams of AI scientists from across the globe building a ‘deep learning project’, once named the ‘Google brain’. This is an extremely powerful tool – one which is constantly learning. For example, the team exposed the ‘brain’ to 10 million YouTube stills. It managed to figure out, entirely on its own, what a ‘cat’ was and identify videos with cats in them. For a system which had no previous conception of the feline race, this was monumental – it had developed its own concept of a cat.
In 2013, Google acquired DeepMind, a London-based AI company. In a system similar to algorithms, DeepMind’s program learned to play Atari games better than a human being by playing millions and analysing strategies – inventing techniques to win that no living being had ever tried before.
Recently, Google has been bringing aspects of its ‘deep learning project’ into search results and other areas like real-time translation. Where this will lead ultimately remains to be seen.
How it affects business
Google is the most used search engine on the planet. There are over 100 billion searches a month, contributing to a 75.2 per cent share of the search engine market. If you run an online business, your success will be directly affected by your rankings in Google. Recent updates have penalised sites with poor mobile optimisation and introduced an AI ‘RankBrain’, which can determine how relevant your site is to a user’s query. With this increasingly machine-led style of search which will eventually be far beyond human input, are all businesses at Google’s mercy?