Game mat

Travel through time on our game mat

Who will reach the finishing line first and help determine the future of AI?

START

01 - Phase 1

In the first phase – after WWII – computers with AI are not yet very smart. Experts give computers lots of instructions, and the computers have to follow these step by step to find a solution. These are what we call “expert systems”.

  • Hop on one leg on your next turn(s) until you have passed square 5.

03 - 1950

‘Can a machine make someone believe that it’s human?’ This is what computer scientist (and LGBTQIA+ icon) Alan Turing wonders. He comes up with the Turing test: by chatting with a human and a computer, you have to find out which is the human and which is the machine. If you get it wrong, the computer has passed the test!

04 - 1955

The American John McCarthy is the first to use the term “artificial intelligence” – or AI for short.

  • AI could also stand for other things, such as “awesome iPhone”. Can you think of three other things? If you succeed, you can move two steps forward.

07 - 1965

Gordon Moore states that the number of electrical components that fit on a computer chip will double every two years. This is known as Moore’s law. What does it mean in everyday terms? That the power of computers will grow greater and greater – enabling AI researchers to make more and more progress over the years.

08 - 1966

ELIZA is one of the very first chatbots. In fact, it’s a very basic ChatGPT. ELIZA imitates a (not very good) psychologist and asks an incredible number of questions.

  • Can you have a conversation which consists only of asking questions? Get someone to ask you three questions and answer each time with a question of your own. If you succeed, move one step forward.

10 - 1974

Start of an “AI winter”. This is a period in which there is less interest in and funding for AI research, and it lasts until about 1980. The reason for it is that people had exaggerated expectations for AI that weren’t met.

  • Move one step back.

12 - 1977

‘A long time ago in a galaxy far, far away...’ the sci-fi film Star Wars is released. And who is the star of the film? R2-D2, a robot with AI. In the film anyway – in real life there was an actor inside the robot.

  • Roll the dice twice. If you roll the same number twice, you can move that number of steps forward.

13 - Phase 2

In the 1980s, AI scientists shift their focus to machine learning: the possibility of getting machines to learn things themselves. This marks the start of AI’s second phase.

14 - 1997

The computer Deep Blue beats the then world chess champion, Garry Kasparov – a bitter defeat! Garry was not amused.

  • Get someone to tell you the best joke they know. You mustn’t laugh. If you succeed, move one step forward. If you laugh, move one step back.

16 - 1998

Scientists create the robot Kismet. It takes the first steps in recognising people’s emotions and imitating them.

  • The person in last place looks very sad. Swap places.

17 - 1999

The company Nvidia brings out a more powerful kind of computer chip: a graphic processing unit or GPU for short. Standard chips or CPUs usually perform calculations one by one, but GPUs perform multiple calculations at the same time. This will prove essential for AI.

  • Move two steps forward.

19 - Phase 3

Since 2010 we have entered a third phase of AI: deep learning. Deep learning has actually been around for a lot longer, but only becomes more useful now due to the enormous improvements in chips. It’s a form of machine learning that works with "neural networks”. And what are they inspired by? The way our brains work!

23 - 2011

‘Siri, how can I cheat in a game?’ Apple’s new AI assistant, Siri, has been able to help you since 2011.

  • But that’s not fair, is it? Move a step back.

26 - 2012

With the help of chips made by Nvidia, an AI system learns to recognise a cat.

  • Imitate an animal without making any noise. Can anyone recognise which animal it is? If so, you can move two steps forward.

30 - 2014

Eugene Goostman is the first chatbot to pass the Turing test (see square 3). Eugene pretends to be a 13-year-old boy from Ukraine. A third of the people who chat with him are convinced that Eugene is human.

  • Swap places with the person in the lead.

34 - 2016

Google’s AI system AlphaGo defeats the world Go champion, Lee Sedol. Go is a difficult game. In chess, there are around 35 possible moves per turn on average. In Go, there are around 250. Lee later realises that he will never be the best again and retires.

  • Skip a turn.

36 - 2017

The first deepfakes appear on the Internet. These are videos or photos that have been edited or manipulated with AI. In 2017, lots of them are pornographic videos in which the women’s faces are often replaced with those of Hollywood stars.

  • That’s misogynistic! Move three steps back.

40 - 2021

Entering a text and having it converted into an image – that’s what you can now do with DALL·E.

  • Have you ever made an image with AI? Yes? Then move two steps forward.

43 - 2022

The company OpenAI unleashes ChatGPT 3.5 on the world. You can talk about anything with this smart chatbot. Or you can get it to help you with your homework.

  • Smart thinking! Move one step forward.

47 - 2024

Sora is the latest product from OpenAI. With this AI tool you can convert short texts into realistic and imaginative videos.

  • Describe a film in three sentences without mentioning its title. Can the other players guess the film? Then you can move one step forward.

48 - 2024

The AI Act, a new law, is approved. Anyone who develops or uses AI technology in Europe has to follow certain rules.

  • From now on, you must always shout ‘hop’ for every square you move forward. If you forget, you have to go back to square 48.

50 - Future

All AI at present is artificial narrow intelligence or “ANI”: it can perform one or a few tasks very quickly. In the future, though, there may also be AI that can do anything a human can: artificial general intelligence or "AGI”. But how will we know when AGI exists? There are various proposals for testing this.

  • Can you come up with a test yourself? If so, tell us about it and move one step forward.

51 - The coffee test

AGI test: A machine can make coffee in a house it doesn’t know, with a coffee machine it doesn’t know either.

  • You can take another step forward.

52 - The modern Turing test

AGI test: A machine is given 100,000 dollars of starting capital and can turn it into 1 million dollars.

  • Another step forward.

53 - The IKEA test

AGI test: A machine can look at the parts and instructions of a flat-packed IKEA product and then assemble the furniture correctly.

58 - Further into the future

Perhaps there will also be really smart AI that outdoes humans in almost every way. That would be artificial superintelligence or “ASI”. If AI becomes smarter than humans, it will start to improve at an incredible speed. This is what we call the “singularity” – an explosion of intelligence.

  • Do you think that’s a good thing or a bad thing? How would ASI treat people? Ask someone this question and wait for their answer.

Well? move one step forward.

Badly? Move one step back.

59 - Further into the future

ASI thinks you’re a cute puppy and takes good care of you.

61 - Further into the future

ASI thinks you’re an annoying ant and destroys you.

  • What a shame! Go back to square 48.

63 - FUTURE

Hurray! You’ve won! Ready to use your knowledge to shape the future of AI? Who knows, you could become the inventor of the latest AI application!

Display cases

In 4 display cases next to the game, you can also discover some cool artefacts.

A. Alan Turing

Inventor of the Turing Test | Cracker of secret Nazi codes | Portrayed on £50 note

  • Need a challenge? Try doing more for humanity than Turing.

B. John McCarthy

Father of AI (and of Susan, Sarah and Timothy)

C. ELIZA

(Incompetent) psychotherapist | Chatbot

  • Okay, maybe not completely incompetent. When the inventor, Joseph Weizenbaum, had the chatbot tested by his secretary, she soon asked him to leave the room. Why? Because the conversation was becoming too personal.

D. R2-D2

Robot with a mind of its own | Fun LEGO construction project for the creators of this exhibition who needed to relax

E. Go

Niet het spel van de wedstrijd Lee Sedol – AlphaGo

  • Because we couldn’t find it.

  • On the 37th move of the game, AlphaGo placed a stone 5 rows deep. This was something a human player would never have done. Lee laughed at first, but then stared at the board for 12 minutes. The move was brilliant, and Lee was staggered that a computer could make such a creative move.

F. DALL·E

Result of the first version of DALL·E |

Prompt: An avocado armchair

G. Coffee machine

Available for AGI test

  • The only instruction the robot gets is: ‘Go into the house and make coffee.’ This means that the robot has to open the door to the house itself, find its way around the house, identify the coffee machine and then use it to make coffee. Latte art is an optional extra. Let us know if you’ve built a robot that could do this.

H. $100,000

Available for AGI test

  • No, this isn’t real money.

I. IKEA instruction booklet

Available for AGI test

  • Assembling an IKEA cabinet is challenging. But it’s even harder to visit IKEA without inhaling 500 Swedish meatballs