#ADE2017: Artificial Intelligence is everywhere

The term Artificial Intelligence is everywhere. It's a hype. Nowadays, every new startup pretends to do something with AI. But the term is nothing new at all. AI is around since the 50's. It took a while, but thanks to the cheap and strong networked computer power we have nowadays, AI is everywhere. And in music too!

For ADE (Amsterdam Dance Event), Paradiso Labs joins forces with Paylogic, curator of the one day event ADE Tech. Warming up to the event, we write a series of blog posts about trends and new technologies and their potential impact on the music industry. Do you want to visit ADE Tech on October 19th? Here's more info about the program.

Computers that make rational decisions

Let's start at the beginning. We won't go into deep detail. So we won't go into all the different definitions people use when describing forms of smart computing. Because in essence, that's what it is when we talk AI or Machine Learning or Deep Learning. To keep things simple, specialists will say that I oversimplify, in this article, I will call them all Artificial Intelligence.

Artificial Intelligence (AI) is the science behind making machines that act and make decisions rationally. (Nipun Ramakrishnan)

So when a Tesla car makes decisions in traffic, AI is involved. When Siri understands you, AI is involved. When Netflix helps you find relevant content, AI is involved. And in music? The best example is Spotify of course. The company is excellent in using smart computing to make personalised playlists. Spotify learns from all of its users and makes recommendations better with every play. There is another definition, that helps:

AI is a subfield of computer science that helps computers make accurate predictions on any type of data. What this means is that instead of explicitly telling a computer how to solve a problem, you show it how it was previously solved and the computer learns on its own on all the steps required for a solution. (Sarim Haq)

Sarim explains it nicely. "For example: let’s say you want to train a computer to recognize a pen from a pencil. Pens and pencils come in a variety of sizes, shapes, colors etc. Previously, you would have hard-coded the computer to test for every single one of those combinations before reaching a decision. Now you can just train it on 50 pens and pencils and it’ll infer, on its own, which attributes differentiate them from each other, for example, a lack of lead tip. This saves engineers hours and hours of coding and reduces the cost of development."

AI is used in face detection, object recognition, speech recognition, hand writing recognition, but as mentioned in recommendation engines of Spotify, Apple and Deezer as well. But there are other examples. Feed a computer with a million hits, let it analyze the tunes and make predictions on future hits. Or let it produce music. So far it is mainly middle of the road or watered down versions of orginal art I've heard, but who knows what the future will bring?

What can Artificial Intelligence mean for a club?

Of course, there won't be a lot of clubs that have AI specialists working for them. But the good news is that we are now entering the phase of the democratisation of AI, affordable platforms where the non techies can make use of. Add data to a system, think what the question is you want an answer on and the computer will do the rest.

For example. An important question a club has is: is the number of tickets an artist is going to sell predictable? Of course, the programmers of Paradiso know a lot and have a good gut feeling. But it is always good to have the stats to support you. It can help you book an artist in the right size venue. It can help you calculate the right price to pay an artist, etcetera.

In this case, what data can you feed an AI machine to learn from? There is a lot. Twitter followers, Instagram followers, Facebook followers, YouTube plays and followers, Shazam searches, radio and tv plays, etcetera. And of course Spotify is a gold mine with plays, monthly listeners, followers, top locations, playlists and related artists.

Where possible it would be great to not only have the number of plays, followers etcetera, but the percentage growing or declining as well.
And the more localised the information can be, the better. Paradiso wants to know how big an artist is in the Netherlands - or even better - in the Amsterdam area.

To all the available external data, Paradiso can add this: amount of tickets sold, price ticket, prediction programmer.

Will it be possible to predict the number of tickets Paradiso is going to sell for each of its concerts? How close are we going to get? It definitely is one of the directions Paradiso is going to explore. We'll keep you posted.

But there are other AI options too. Most services have an API that makes it possible to use the strength of their tech infrastructure. Spotify, for example. We at Paradiso use their recommendation engine in our site. When we combine your profile on Spotify with the gigs we have lined up, we can make personalised recommendations for concerts to go to. So we use external AI power to make our site better. Every club can do that too.