2017 was said to be ‘the year of artificial intelligence’, thanks to the many leaps and bounds made in the technology. Of course, in the same year, the hype-mill for AI went into overdrive. Behind all this hype, there are some really great applications in AI analytics that you can take advantage of now, not in the year 3000 – and it’s not all robots, sci-fi, and Black Mirror either.
Firstly, we should really discuss the difference between machine learning and AI. In some circles, it’s become the norm to use these two terms interchangeably. But that’s a mistake.
Machine learning is AI, but not all AI is machine learning. Machine learning is a subset of AI and it describes how computers can learn from data sets. Importantly, the data sets are limited, restricting the AI to just a few different functions (based on what data it has been trained on). So, an AI trained with images of people’s faces will rule at facial recognition, but fall flat with voice recognition.
AI is designed to mimic human intelligence. Instead of just a few data sets, a computer is given access to a heap of different data and just left to learn for itself. It means that a computer will know about many different things, and not just restricted to one or two tasks. Most of the AI you see around us today, that’s machine learning. ‘True’ AI is still a little way off…but we’re getting there.
Ever used a debit or credit card? You’re interacting with AI. When you spend your hard-earned cash, machine learning is busily detecting fraudulent transactions. It also helps financial institutions manage risk when approving loans, credit cards, and mortgages. The AI analytics determine your usual spending habits, income, and credit history, and from this, it can tell whether you’re a safe bet (or not) for one of their products.
Of course, AI analytics can facilitate decision making, but on many occasions, I believe it mustn’t make those decisions itself. There needs to be a degree of human oversight, especially when something as life-changing as a mortgage is being decided. Banks need to balance the risk of providing their product with the need to give a good service and access to products.
Without that human intervention, it creates a risk of AI becoming black box (in which it’s built to reach a certain outcome, and then set loose to make decisions, such as trades on the stock market). That’s not ideal, as it can lead to biased decision making that isn’t instantly recognisable as such.
There’s a good example in Weapons of Math Destruction, where an algorithm began automatically approving white male college applicants at a much higher rate than their black counterparts. Obviously, this isn’t the objective of deploying AI in your business, so more thought and guidance is required.
Make sure you always know (and document) how your AI comes to its outputs and any decisions. Also, make sure a human always has a handle on things. GDPR now requires as much – it’s a legal requirement where personal information is involved.
We all dream of inbox zero. Thanks to machine learning, that’s a much more possible task. Spammers are using increasingly sophisticated ways to break into our inboxes, meaning that traditional spam filtering techniques are about as effective as a broken umbrella. Machine learning keeps up with the latest spam tricks, and works to keep the spam out of our bulging inboxes.
Conversely, there’s also AI that can answer your emails for you. Okay, so it’s not going to tersely reply to the fifth round-robin that Kevin sent this week, but it can provide some standard messages to help you focus on other work. I’m fond of the x.ai programme, which I use to save time when booking meetings. You’d be surprised by how much of your working day is taken up with these little activities.
Therein lies a small, but mighty function of AI. In taking over our menial tasks, machine learning allows us to work on higher-level strategic activities. It’s probably the first experience your employees will have with AI at work, and an easy way for your business to experiment with integrating AI analytics tools. Plus, it gets employees used to AI without imagining the Terminator as their future co-worker.
Specialised AI is being used in the legal, marketing and HR fields. In legal, it is helping lawyers sift through paperwork faster than ever before. This means mergers and acquisitions are completed more quickly (meaning a better deal on the table), court case outcomes are predicted with greater accuracy, and junior lawyers no longer have to burn the midnight oil highlighting cases for their seniors.
In marketing, machine learning is helping companies craft super personalised and targeted communications. Leading to less irrelevant marketing for everyone, and saved revenue (with greater returns) for businesses.
And in HR, AI analytics is proving to be a great ally. It can categorise CVs, saving a lot of time for harried HR departments and assist with interviewing candidates. Companies like Headstart AI are offering companies the ability to personality test candidates, for instance.
The uses of AI analytics are incredibly varied, but there is some common ground. For one, they all need to start somewhere; data underpins everything.
It’s no surprise that the companies leading the way with AI (Google, Amazon and Apple) have a mountain of data. The battle to reign supreme over AI analytics is going to be won by data.
Google has a huge amount of data at its disposal. There’s not only all our search data, but also data from Gmail, Adwords, and now from Google Home and Pixel. It illustrates perfectly how valuable data is today. If your business isn’t using its data, you’re wasting a precious resource.
Of course, you do need a plan to make the most of your data. If you plan on implementing AI analytics, you’ll also need to prepare your data strategy so that AI can work with it. A growing challenge that we’re seeing time and time again is that organisations often aren’t ready for AI. Is yours?
There’s also something to be said about the way AI analytics are being used currently. Yes, it’s not yet ‘true’ AI, and is rather limited in its current form. However, even when ‘true’ AI is developed, it won’t take over every job on the planet – AI will work with us, to make us better. That can be something as simple as clearing out junk emails, cleaning up a database, or capturing the key points in a document, all the way to advanced AI that supports soldiers in the battlefield and drives our cars.
Just think about the difference that Amazon’s Alexa has made to many homes. That kind of consumer-focused AI is going to have just as much (if not even more) impact in our workplaces. We’re going to be entering a new era of intelligence, one where our own is supplemented by AI. It’ll make us work more efficiently, faster, and with greater accuracy.
Change is difficult, and AI is bringing a huge change to our lives. We’re just about getting used to having machine learning in our homes and on our phones, but soon it’ll be in every aspect of our day.
For businesses, there’s a degree of future-proofing that you can do. You can keep up to date with the latest AI analytics tools available and under development (CognitionX are doing great work in this space to help bring clarity to it). That way, you can determine whether the tool will help your staff at all.
Data is like gold dust in the age of AI, so make sure you take advantage of it. The first thing you’ll need to do is create a data strategy, as your entire use of data runs off of that. The second thing you’ll probably have to do is hire in someone data-savvy – that means a Chief Data Officer, or similar to help lead the charge. They’ll be able to steer you right when using your business’ data. You might also consider a Chief AI Officer, but I’d argue they aren’t completely necessary. However, you will need some AI and machine learning expertise accessible to your business, to use it to its full potential.
Start with the opportunity you have or the problem you are trying to solve, and work out whether AI analytics is the right solution. I’d definitely advocate experimenting with some AI now, before the market gains full maturity and everyone hitches a ride on the AI bandwagon. When everyone else is jostling to get on, you’ll be miles ahead.
For more insights into building a data strategy, make sure to download our Data Strategy whitepaper.