Artificial Intelligence: What exactly is it and does it contain any E numbers?

Hello hello!  So, after 4 years of my personal blog, and 50,000 followers later, I been asked by quite a few of you (thanks Mum!), to share my thoughts on the wonderful world of Analytics that I’ve been living in for the last 25 years, with a focus on where the industry is going, and explaining in ‘Min Language’ all those techie phases that you might may have heard in that last meeting you had with the data scientist you’ve just hired (as that’s what your competitors did), or you overheard the topic being discussed by the youngsters on the H12 to Harrow on your way home when they weren’t on Snapchat.

Let’s tick them off: AI, EI, Machine Learning, Neural Nets, NLP, NLG, RPA, VR, AR, MFI, that kind of thing. Ok, I made the last one up, but you get the gist. “EI”…didn’t get that one? That’s Emotional Intelligence. More about that later…

I believe we are entering the most exciting times of mankind right now, as we finally are at a stage where the stars have aligned around us having the necessary raw computing power, access to large amounts of storage in the shape of that fluffy cloud and finally access to data.   

I thought I’d kick off a topic that Hollywood and Bollywood has been bringing to our screens since the invention of the silver screen. That topic is ‘Robots: Friend or Foe’. I’m sure you will all have a view as if you are in the Terminator camp (the first one when he was a bit naughty) or iRobot where our silver friendly humanoid was on the side of Will Smith. There is a great debate by leaders such as Bill Gates, Elon Musk (my hero!) and Steven Hawking who say we need to be really careful that the current AI path we are on could lead to the biggest risk, that we face as a civilisation,  where Skynet is a reality and we are ruled by computers. Yes folks, a bigger risk for mankind than the effect on our brains of us all secretly watching that last season of Love Island, though we would never really admit it to our friends.   As homework for this blog, I watched a BBC programme that was on last week called Hyper Evolution: Rise of the Robots.

It was a fascinating watch to see how over the last 100 years the robots have got much more intelligent, having taken over many of the tasks we used to do like making cars, walking up steps, and playing the piano…For me, I was most interested in how through the use of Artificial Intelligence, robots have started to learn all by themselves, by using previous conversations and continuous feedback to improve, whereby they know the difference between a tea cup and an orange. (hint, you can’t dunk your biscuit in the latter).

So what do we actually mean by the phrase Artificial Intelligence, and is that different to Machine Learning and what about Deep Learning? How deep is deep? Well, I’m glad you’ve asked.  

Artificial Intelligence (AI) is a branch of computer science that aims to create intelligent machines that work and react like humans. For example, the computer has certain traits such as Knowledge, Reasoning, Problem Solving, Perception, Learning and Planning. All the lovely things we do without really thinking about, like that time when the driver of the Met Line tells you the tube is now terminating at Wembley Park as he’s going on strike and you now have to work out how to get to Moorgate!

So a subset of AI is Machine Learning. (It’s not something to use interchangeably, unless you want the geeks at the next Hackathon you attend to roll their eyes at you). In essence, Machine Learning is a technique where the machine is ‘trained’ using large amounts of data and algorithms that gives it the ability to learn how to perform the task. They do this by detecting patterns in existing data, identify similar patterns in future data and then make data driven predictions. Still don’t know what I’m talking about? Let’s try bringing it to life. The next time you are on Google and you misspell a word (very likely these days thanks to reliance on autocorrect), and you type ‘Ficebook’ as it’s lunchtime and so that’s fine as you are covered under Human Rights legislation. The machine learning algorithm, will suggest that you probably meant ‘Facebook’. It even works if you type ‘FB’. Smart huh.

Ready for another example? This time let’s have a look at one of my favourite sites that I love to show to our clients when they come and visit our lab, here at Concentra (www.concentra.co.uk).

https://www.captionbot.ai/  Here, it’s using some more powers within the AI family, specifically computer vision and natural language to describe what’s in a picture. Now for humans, that’s easy peasy, but up until now, really hard for a computer to do. The website uses an image library from all the pictures stored on Bing as it’s frame of reference as well as a super smart emotion algorithm to describe if a human in a picture is laughing because they are still remembering that MFI joke.  The site has some sample photos like the one below, and what’s really impressive is that by using these natural language generation techniques, it stiches a sentence together like a human would by understanding the different objects in the picture to work out what the context could be.  What’s great is that the system is continually getting better by the users giving  a ‘how did I do’ score.  I remember last year,  I tried testing the system with a picture of Donald Trump (before he was President), and at the time it thought it was Louis van Gaal, the Ex Man United football manager. So I gave the machine one star and no dinner and now as you can see below, it’s much more accurate…

Woah, I’ve just realised that I’ve written quite a bit already today on this super topic and as it’s probably time for you to get off the tube now, or go to that meeting, why don’t we leave it there, and we can pick up again next week to continue this topic.

Please let me know if there are any other topics that you would like me to demystify like ‘Are we really living in the matrix’, or ‘Will my kids need to bother how to drive’ or ‘Is your name really Min or is that some kind of geeky statistical inside joke’…that kind of thing.

Right, let’s leave it there, and I wish you a lovely sleep,

A small man called Min