Why Artificial Intelligence, Machine learning, Deep Learning, and cats should matter to your property business. By, askporter

Artificial Intelligence (AI) is one of the tech world’s favourite buzzwords – right up there with blockchain. I’m confident that somewhere on Amazon you’ll be able to buy an AI-assisted toaster that learns how you like your toast and logs each toasted slice to the blockchain for ‘toastal transparency’.

Given this hype, it’s understandable that software developers are desperate to cash in on ‘AI’. With a unique mix of frustration, bemusement and at times mild enjoyment, I’ve sat through pitches where people say “oh and we’re looking to integrate AI in the next 6 months”. Hopefully, when you get to the end of this article, you’ll have a better idea of which AI companies you should trust, which you should marry, which you should avoid and what AI actually means.

While some might be sceptical given all this talk, Artificial Intelligence is not the typical hyped innovation we’ve come to expect from the tech industry. It is without a doubt, right up there with the internet, the Walkman, and the wheel. However, it’s a little more difficult to understand and explain than Walkman or the wheel.

The reason AI has come to the fore in recent years is that data processing has become cheaper, faster, and a whole lot more powerful than we’ve ever seen before. Data itself is also in abundance, and with almost infinite storage at our fingertips, there seems no end to what can be achieved with the deployment of the right code. Just take self-driving car technology -, driving a million cars used to take a million humans, now it can be done with one deployment of code.

Acronyms and buzzwords can cause discussions around AI to get very muddied, very quickly. The main cause being misunderstanding the nuances of Artificial Intelligence, Machine Learning, and Deep Learning, and why that distinction matters.

The easiest way to understand is to remember that all deep learning is a type of machine learning, and both fall under the umbrella term of artificial intelligence. An often used visualisation of this is to draw – or imagine – three concentric circles. The largest circle is AI. Inside that is the smaller circle of machine learning, and inside that the smallest circle, deep learning.

Ernest Davis, a professor of computer science at New York University gives a good example that illustrates the fundamentals of AI – say you want a computer to know how to cross a road, for conventional programming you would give it a very precise set of rules, telling it how to look left and right, wait for cars, use pedestrian crossings, etc., and then let it go. With machine learning, you’d instead show it 10,000 videos of someone crossing the road safely and 10,000 videos of someone getting hit by a car, and then let it do its thing.

  1. Artificial Intelligence

    Without going too deep into every aspect, Artificial Intelligence has been around for a very long time as a research field. In fact, it has been explored in varying degrees since the 50’s. Until recently, a great deal of AI discussion was dismissed as the fantasies of sci-fi gearheads. In a nutshell, AI is a broad, all-encompassing term that is used to describe anything other than humans that have been programmed to exhibit or perform a human-type ability.

  2. Machine Learning

    Machine learning is a type of AI. It is the more specific practice of using an algorithm to parse and input data and make determinations and predictions relating to the real world. Essentially any time a machine is trained by its experience can be considered machine learning in action. Over the past couple of decades, people have tried all sorts of different methods to try to train ‘machines’, such as reinforcement learning, which is like toilet training a cat, giving rewards every time the cat uses its litter. Then there are genetic algorithms, which are more like taking 2 cats and letting them compete to be the best cat across a series of tasks, a little like natural selection. These are both types of ‘machine learning’.

  3. Deep Learning

    Ah! And now Deep learning. Deep learning is a complex method of machine learning. To return to the cat analogy, deep learning is like having a load of cats living in a tribal hierarchy. The lowest cats on the hierarchy perform basic tasks and report back through the hierarchy, with each hierarchical layer drawing more insight than the previous one, until the ‘top cat’ has a clear picture of what’s going on. Basically at some point, the tribe of cats will go, “hey, we’ve seen this kind of pattern before. It’s probably the same thing.” (I spent a while working out that cat analogy and it’s not perfect, but hopefully you get the idea).This process is called Deep Learning and is used to perform more complex machine learning, such as language, speech or image recognition.

Why does it matter?

Machine Learning is often incorrectly assumed to be the same as deep learning. The two have become synonymous, to the detriment of business owners who might believe their latest and greatest software acquisition is much more powerful than it actually is.

At AskPorter we leverage the latest tools in Machine (and yes in many cases deep) learning to engage with our client’s customers and perform tasks. In fact, we’ve put these tools right at the heart of our software, so with every interaction (even when people are communicating with other people) our software delivers an increasingly better service and drives efficiencies at an ever-accelerating pace.

I will finish by suggesting that like most things in life, when it comes to AI it’s not what you have, but what you do with it that counts. These days Microsoft, Google and Amazon all have great off-the-shelf machine and deep learning tools. So when shopping for a provider, ask what data-set they have trained their software on, the quality of that data and what models they’ve built from it.

In case you were wondering, at AskPorter we have over 70,000 categorised and analysed interactions from which we’ve developed hundreds of property specific models and workflows.