AI: you can’t seem to escape the talk around how computers and machine learning will take over all processes in business. And not participating in it, is like becoming a dinosaur. But as with any new technology, business should approach this with a healthy dose of scepticism. The world is not going to change tomorrow, and early adoption is not always good for a company. There are plenty of seminars and conferences you could go to that will discuss AI, but sometimes you will leave those with more questions than answers. Here is a quick exploration of what AI is, what it could mean for your business and what a good moment would be to step into the field.
What is AI?
First of all, we will need to tackle the concept of AI versus machine learning as these terms seem to be used interchangeably. Artificial intelligence (AI) and machine learning (ML) are related, but they are not the same. AI is the idea that a machine can perform tasks that are ‘intelligent’. This breaks away from the notion that you have to pre-program every single possibility beforehand. So whenever a machine is able to solve problems based on a set of stipulated rules, that’s what we call it AI.
Generally speaking, there are 2 types of AI, one is narrow, and the other is generalised. Narrow AI is, as the name gives away, a system that is designed for very specific tasks. Popular examples are virtual assistants such as Siri. General AI is, as you would suspect, more complex and can, theoretically, handle a broader set of instructions and output, but due to the complexity are very costly to develop and maintain. An excellent example of this is IBM Watson.
AI And Machine Learning VS Deep Learning
Machine Learning (ML) is a method in which AI can be established. What people often get wrong is to assume that ML is the only way to develop AI. It probably is the most successful and popular way to develop AI. Machine learning relies on training a machine to find ‘logical’ relations between data points you feed it. An example could be feeding the machine shopping receipts of the same store over and over. The challenge could be to blank out some of the prices at one point and see if the machine can figure out what these were supposed to be.
You will also have heard the term deep learning as well, which also is a form of machine learning. Most people would say it’s the evolution of machine learning. Where with machine learning you are getting a system to ‘learn’ based on the data you feed it, deep learning is an attempt to mimic the workings of a human brain, by utilising the short cuts human brains employ. When humans learn something new, they will compare it to something they already know. This type of categorisation helps the brain to process and understand new concepts quicker. So instead of creating task-based algorithms, the aim is to learn what the data represents. This type of higher analysis obviously requires additional computing power. This is where we tend to talk about how this data would be stored, in deep neural networks.
An excellent way to think about the differences between AI, ML and deep learning is to say AI encompasses everything. Within AI you have the concept of ML and within ML lies the concept of deep learning. That does not mean that in-depth knowledge is everything to ML, or that ML is everything to AI. Now we have established a good understanding of AI, ML and deep learning, it’s time to look at some good examples of AI. Some of these you will already have deployed.
You’re Probably Using AI Without Realising It
As mentioned, Siri is a form of AI that people regularly. As with Siri, you can also say Alexa, Jasper (formally known as Jarvis), Cortana and Google Assistant are forms of AI. An extension of that is using voice-to-text features. So if you have ever dictated a letter or even a quick text-message, you will have used an AI. Text-based AI is enormous as it has to take into account any local characteristics of the spoken language. Not something you can program every probability for. Hence AI plays a role. Codifying language brings its own challenges (and rewards) and transcends just language.
Another form of AI that you would have benefitted from, possibly without even knowing it, are email spam filters. As would-be spammers continuously change their tactics, so does the protection need to update itself as well. This is a funny one as it also could be very well the case that on the other side (the spammers) they are employing advanced tools themselves, and maybe even AI. In the same vein, AI can be linked to CCTV, being able to analyse faces and microexpressions to assess someone’s state of mind and find potential risks.
You might have noticed that you are able to search photos now based on a simple text search, you can try this if you have an iPhone. This is all down to AI being able to recognise objects in photos, even on your own collection. More playful is, for example, Snapchat filters. Image recognition is pivotal in growing AI. The combination of imagery, spoken and/or written language gives a fairer representation of the data points humans have to deal with and therefore will bring AI truly closer to human intelligence.
Also in the world of translation, people are hard at work to employ AI. Although you might have some misgivings around Google Translate and some of the smart translations apps and hardware, it is a space to watch. People are still working on devices that allow you to travel the world without knowing any other language than our own and be perfectly fine in understanding people around you and making yourself understood.
More flashy examples of AI can be found in robotics, even to the extent of seeing humanlike robots that are trained to looks like humans (but always miss the mark). The combination of chatbots and having human-like robots is powering a very common ambition of companies to replace front-of-house services and enhance customer service quality (and availability) for users. It wouldn’t be all too surprising that the consumer banking world and hospitality sectors are leading the charge here.
It’s Not Just About Talking Tech
Also under the robotics category: robotic vacuums. Amazing little devices that can keep your house tidy. It’s one of those gadgets you want to have, regardless of whatever AI is powering it. There seems to be some promising consumer success to this end, with robotics vacuum/floor mops and lawn mowers fairly popular in sales. As the hard and software improve, the novelty factor will make way for the ‘must have’ label. There also are some more experimental AI proofs-of-concept being developed such as a robotics chef and personal butler, but it may take a few leaps in physical capability before this can become more mainstream.
That covers a variety of some of the more popular AI uses, for businesses AI can play a role as well. Most of these applications are around finding efficiency or reducing wastage in the workplace.
Take investing for example. Especially with the rise of fintech, a lot of solutions out there are powered by AI. For example, AI that helps you decide and make decisions around your investment portfolio. AI that enables you to save or anticipate future expenses. AI that coaches you getting your spending habits under control. AI is ideally suited to learn and understand numbers quickly, and these more personal finance solutions are only the start.
Even with something as mundane as invoice matching AI can play a role. If AI can help you take care of the back office processes of matching bills to POs, you can increase your PO spend. This could be a game-changer for many companies being able to maximise the opportunity they have been given.
Also for company travel, AI can play a huge role. Optimising routes and making travel arrangements can be quite a complex task, especially if done by humans. AI can perfect this and provide the ‘best’ answer in a fraction of the time needed for a human to do so. As with robots playing a role in front-of-house functions for the hospitality sector, so will AI play a transformational role in these sort of optimisation tasks.
For most businesses, AI will play a massive role in terms of marketing and communication with clients. With client touch points becoming less linear and more multi-faceted, humans can no longer keep up staying on top of it. AI will play a huge role in making these micro-decisions in marketing processes.
Regardless if you are an AI aficionado or a sceptic, it’s hard to deny that AI isn’t already here and becoming more useful every day. Finding the balance, however, is critical. If it feels gimmicky, it might not be the right time, but if you can quantify (and justify) the cost versus benefits, it might just be a simple (and logical) business decision.