Preparation

Zur Vorbereitung ist lediglich ein PC und eine Internetverbindung notwendig. Alternativ kann diese Übung auch an einem Tablet oder Smartphone stattfinden.

Das Video "AlphaGo" kann auch zur Vorbereitung individuell (zu Hause) angesehen werden.

Method / Process description

In this exercise, the topic of "artificial intelligence" is discussed using two examples. There are few guidelines in this exercise, rather the aim is to initiate a discussion. 

AlphaGo

Go is a strategic board game with 19 columns and 19 rows. Two opponents must alternately place white and black pieces and try to go around the opponent's pieces. Once each player has moved once, there are already 130,000 possibilities - compared to only 400 in chess.

In 2015, the company Google DeepMind developed AlphaGo, an artificial intelligence programme to play Go.

In 2016, AlpahGo played against Lee Seldon, one of the world's best professional players and then world champion, and won.

Search the internet for the video "AlphaGo" or "AlphaGo documentary". Watch the documentary and try to answer the following questions:

1. Why did the developers choose the game Go instead of, for example, chess? What is the main difference between the two games?
2. Why were the viewers, press and Lee Seldon so confident that AlphaGo could not win?
3. Why were the spectators so surprised when AlphaGo made a strange move in the second game?
4. Why was Lee Seldon so confused when he played against a non-human opponent?
5. What did Lee Seldon learn from playing against the machine?

Answers to stimulate discussion: 

1. There are more combinations of pieces on the Go board than there are atoms in the universe. It is impossible to do a statistical analysis and combinations of all possible solutions; therefore, Go is a game that cannot be played by brute force, but a machine must try to think
2. Because Go requires a strategy, something only a human can have
3. Because a human would never make such a move. It was something whose meaning no one could understand, at least not at that moment
4. Because he felt like he was playing against himself. When you play against a person, you try to perceive his/her emotions and take them into account as part of your strategy ("exchange of feelings"). Now this information was nil. Lee Sheldon was only playing against himself ("more questions about himself").
5. New ways of thinking and approaching the game that he had never thought of before. New ways and new possibilities ("What surprised me most was that AlphaGo showed us that moves that people might have thought were creative were actually conventional").

MoralMachine

A lot of research is currently being done on autonomous cars. But one of the biggest challenges here lies predominantly in the legal and ethical issues.

How should an autonomous vehicle behave if an accident can no longer be avoided? If an autonomous car is driving along a road and suddenly a child crosses the road, what should the car do if it has to decide either to turn the steering wheel and hit an oncoming car or run over the pedestrian? 

People confronted with such a situation act instinctively and for effect, not thinking carefully. A self-driving car, however, acts on the basis of programmed rules, i.e. decisions made in advance by the developers.

Task:

• Open the website https://www.moralmachine.net/
• Let the learners play through the scenarios alone or in groups 

Questions for discussion: 

• How easy or difficult was it for you to evaluate each situation? 
• Do you think it is possible to find conclusive rules for all situations? 
• What would be your personal approach to finding a solution in this area? 

Note: It will not be possible to find a conclusive answer here. Therefore, it is important to stop the discussion after some time. A good exit is e.g. to create a list of PROs and CONTRAs of autonomous cars. 

Further tasks: 

• Find out in which places there are already cars without drivers nowadays.
• Would you personally get into a car that drives automatically (i.e. without a driver)?

Download material

-

References

-