Who would have guessed; AI can be fun…

20 Dec 2018

One of the great things about working for an AI start-up is that it’s always a solid conversation starter at parties. Everyone has some idea of how AI is revolutionising their industry; everyone has some idea of what they think it actually does.

It can, however, lead to some confusing interactions when people realise that I’m not quite as ‘tech-bro’ literate as they might have anticipated. At Discover, we’re all about how AI can help people ‘be more human’, not overtaken by machines. And there’s nothing more human than having a bit of fun at Christmas!

So if you want AI to do something useful and brighten up your Christmas, and your AI party conversation, then try these surprisingly entertaining AI based games below.

(If you want to hear a bit more about how we at Discover are using AI to help people ‘be more human’ by accelerating expertise to discover brand growth opportunities, check out our recent White Paper).

 Step 1 – Machine Pictionary


This is a great little exercise where you’re given a word to draw (apple, bicycle, wristwatch etc) and the machine tries to guess what it is.

What’s fascinating about this is that it has learnt based on what other people have drawn in response to that same word – so you playing the game makes the game work better for the next person.

It’s fun when it works, but far more interesting in the places that it failed, as demonstrated by my attempt to draw a raccoon: 

Machine learning's effect on human processes Machine pictionary

The artist in me is offended by the machine’s clear lack of recognition at my amazing drawing. My tech side can look at the examples and see why it was confused. The researcher in me is fascinated by the fact that most people, when asked to draw a raccoon, draw it from the front on… 

The relationship between AI and human processes

The other fun thing is how quickly it can learn: I only had to draw this for it to guess harp, based on much fuller examples 

Machine pictionary Machine pictionary

(The semiotician in me has so much to say about visual shorthands I don’t even know where to start) 

Step 2: Automated Art


As you might have spotted from the previous round, my drawing skills may have left me at a bit of a disadvantage. Luckily, there’s now a way for machine learning to help me with this too – it’s taken the logic of the previous game and pushed it further.

In the above tool, if I draw a snail as below, it will say ‘did you mean’ and give me a bunch of options – a kind of visual autocorrect.

How Ai accelerates human thinking


And I did mean! Now my snail looks like this:

How Ai accelerates human thinking

This is in no way useful for my job, but still pretty cool!

 Step 3: Making Music


I have even less musical ability than I do drawing ability, which has saddened me because I’ll never be in one of those musicals where people start making sounds and it turns into a song.

Or WILL I...?!

No, but this site is pretty cool – it takes naturally occurring sounds and turns them into percussion.

It’s then rendered visually here: each dot is a different sound.

Making music with machine learning

They fed the sounds into the machine, and the machine itself worked out which were similar and which were different and clustered them accordingly.

I think what’s fascinating about this is the idea that we’d struggle in words to articulate the difference between the sound of putting a bottle down on a wooden table and on a cork table and how different this is from a metal table. The machine can not only identify these differences, but also provide a visual frame of reference to make it easier to sort through this information, which then allows us to leverage this insight. 

Step 4: Make Your Own Fun

Whilst I really enjoy playing around with different types of machine learning, for me what’s exciting is not just what it does, but what we can do with it. I can’t stop myself from trying to draw conclusions from the types of data people are inputting, or thinking about the different ways of communicating this type of technology is opening up to us.

At Discover.ai, we apply this logic to language and images, using AI to throw up the most interesting content and make speedier connections, for a newer type of research. You might use it to compose the next genre of music. You might use it to do something no one has thought of yet. That's where the fun starts.

Download our white-paper to discover more about how AI can help us be more human.

Download whitepaper

Written by Hannah Marcus
With a background in semiotics and a newly-acquired MA in Cultural Studies, Hannah is almost too excited by overlaps in technology and feminism. Her hobbies include ballroom dancing and talking about her period.

More from our blog

A Cringe World

  We all know how cringe feels, but what does cringe really mean? What in our online worlds can...

Stories From the Edge

There has been a well documented impact on financial behaviours due to the pandemic and lockdown...

Dave the Robo-Cleaner

We bought a Dyson robo-cleaner recently and unsurprisingly it’s proved to be a really useful...

Fandoms - Freaks and Geeks

While fandoms and subcultures are by no means new phenomena, social media and online culture have...

Alternative Diets

Over the last few months, we’ve been exploring the world of ‘Alternative Diets’, and why people...

We are Aura Innovation Award Finalists

We're incredibly proud that our NeedDriver Mapping methodology has been shortlisted for the...

Contemporary Pleasure - Premium Alcohol

The landscape of pleasure; where and how we find moments of joy and experience pleasure is ever...

The Future of Immersiveness

  The pandemic has turned everything upside down in the past year. During the crisis, people were...

Key trends in sustainability

  Over the last 3 years, we’ve worked on over 50 projects for clients, where the central question...

100 projects since lockdown...

In our latest report we have taken a step back to stop and think.. to reflect on these last months...

We are MRS Award Finalists for our Healthcare Research

We're incredibly proud of a recent piece of work completed for the Women's Health Tech Hive,...

Buckle up

by Cathy Maisano   In 1886, Carl Benz patented the first three-wheeled automobile. A couple of...