AI Questions

Development of AI
1

Computing (Higher): Artificial Intelligence: Exam style questions

Modern computers can now demonstrate many aspects of intelligence. Creativity and language processing are two aspects of intelligence which may be included as part of an artificial intelligence system. a b Name and describe one other aspect of artificial intelligence which is currently being developed. Describe two reasons why creativity is difficult to include as part of an artificial intelligence system. A human tester communicates with one human and one computer using remote terminals. The tester is to identify which terminal is connected to the human and which is connected to the computer. a b c i ii State the name of this test. State one limitation of this test. State one strategy that the human tester could use to differentiate between the human and the computer. Explain why the strategy suggested in (c)(i) helps the tester to tell the difference between the human and the computer. An annual competition, based on the Turing Test, is held to find the best example of current artificial intelligence. a b Define what is meant by artificial intelligence. Some people believe that we might never achieve Artificial Intelligence. Do you agree with this point of view? Justify your answer. Natural language processing is one area of artificial intelligence that will enable many more people to use computers. a b Explain how faster processors and more memory have improved the performance of modern chatterbots. Some modern game programs learn how their human opponent is playing and work out strategies to respond. Explain how the availability of parallel processing might aid the development of game playing programs. Computer games have been available since the 1960s. a i ii Explain how the use of parallel processors has aided the performance of computer games. Other than parallel processing, state two other advances in hardware which have led to improvements in the performance of computer games. 2 2 2 2 2 1 1 1 1 1 1 1 2

iii Explain how each of these advances has led to this improvement in performance

Computing Department: Saint Kentigern's Academy

Computing (Higher): Artificial Intelligence: Exam style questions Artificial Neural Networks 6 A building society lends money to people so that they can buy a house. A house valuation depends on factors such as: location of the house given by postcode. number of rooms The building society has access to all this information and accurate valuations of a large number of houses. The building society decides to use this data to create an artificial neural system (neural network) to value houses. a b i ii Give two reasons why this problem is suitable for the creation of a neural network (artificial neural system) A strength of a "neural network" is its ability to learn. Describe the structure of a neural network. Describe how a neural network for valuing houses would be trained. 3 2 2

Computer Vision (and ANS) 7 An artificial Intelligence system is trying to identify plants from pictures of flowers presented to it. This is one of the images it is given to test the system. a b c Two stages in computer vision are object recognition and image understanding. Describe the stages using the above image to illustrate your answer. Give two reasons, other than the colour of the flower, why the artificial intelligence system might have difficulty in recognising a flower. The plant identification software uses an artificial neural system. Links, weights and layers are components of an artificial neural system. Describe the parts played by each of these components in the learning process of the neural net. 3 4 2

Computing Department: Saint Kentigern's Academy

Computing (Higher): Artificial Intelligence: Exam style questions Expert Systems 8 A mobile phone shop uses an expert system to advise customers which phone is best for them. It is essential that the user interface in such a system offers justification facilities a b c d i ii 9 a b i ii Explain what justification facilities do. Explain why these are important to the customer. Explain two problems that the customer might experience when using this expert system. What is the role of the sales staff at the mobile phone shop during the testing stage of the development process? A neural network could also be used to advise the customer. Explain one difference between a neural network and an expert system. Explain how the neural network could be trained. A bicycle manufacturer is developing an expert system to advise customers on their bicycle purchase. State one reason why the bicycle manufacturers domain is suitable. The following paragraph contains some of the information for the proposed expert system. Racing and mountain are two types of bicycle. Racing bicycles have drop handle bars. Hardtails and full-suspension are two types of mountain bicycle. Draw a semantic net to represent this information. An expert system can be created using an expert system shell. i ii d Name and describe one component of an expert system shell. State the component that must be added to an expert system shell to create an expert system. Other than faster development time, state one advantage of creating an expert system using an expert system shell rather than a declarative language. 2 1 1 1 4 2 3 2 2 2 1

Computer Vision 10 a b Mobile robots use computer vision systems when carrying components around a factory. Image acquisition and signal processing are the first two of the five stages of computer vision. What are the last three stages. State two difficulties that a vision system on a mobile robot may have in interpreting a new layout of machinery in a factory. The factory has recently replaced dumb robots with intelligent robots. State two reasons why they may have done this. i ii Describe one legal implication which may arise from the increasing use of intelligent robots. How can this legal implication be addressed by the robot manufacturer? 3 2

c d

2 1 1

Computing Department: Saint Kentigern's Academy

Computing (Higher): Artificial Intelligence: Exam style questions Natural Language Processing 11 a b i ii Natural Language Processing (NLP) involves creating software that uses language in a similar way to people. State two common applications of NLP. NLP has to deal with ambiguity of meaning in sentences. Describe what is meant by ambiguity of meaning in a sentence. The speech recognition stage of NLP correctly recognised the following sentences. Sentence 1: The man threw a bottle at the window and broke it. Sentence 2: The restaurant was full of international food specialists. State one example of ambiguity from each sentence. 1 2 2

c d e i ii

Name the stage of NLP that would attempt to resolve ambiguity in a sentence. Name the stage that takes place after your answer to (c) Explain why it is important to deal with ambiguity prior to this stage. Regional accents or pronunciations, as well as ambiguity, can cause problems for NLP. State two other problems for NLP, using an example to illustrate each answer.

1 1 1 4

12 a i ii ii b c

Natural language processing is one area of artificial intelligence that will enable many more people to use computers. Ambiguity of meaning and similar sounding words both cause problems to developers of natural language processing. Name one other type of problem, generated by everyday language, that developers face. Give an example to show how your answer to (i) may cause the developer a problem. A school pupil uses an automatic translator to help with his Spanish homework. Describe two problems that might be associated with using this translator. Eliza was an early natural language application. Explain how Eliza would select a response to the following user input I find my homework hard. Explain how faster processors and more memory have improved the performance of modern chatterbots. d Some modern game programs learn how their human opponent is playing and work out strategies to respond. Explain how the availability of parallel processing might aid the development of game playing programs. 1 1 2 2 2 2 2

Computing Department: Saint Kentigern's Academy

Computing (Higher): Artificial Intelligence: Exam style questions 13 NSHRDLU is a program which uses natural language processing (NLP) to manipulate blocks of various colours, shapes and sizes. An initial setup of the objects is shown below.

SHRDLU understands commands such as: put the grey cylinder on top of the black block put the white cube in front of the grey block

a i ii b

Here are two commands from a SHRDLU dialogue. What is the problem with each command? put it on its side. put the pyramid on the grey block SHRDLU will accept the command grasp the grey block Explain why SHRDLU might not understand the following command: grip the grey block 1 1 1

c d e

State one reason why there are more problems with natural language processing than with more formal programming languages. In speech recognition software, there are four main stages of NLP. The first is speech recognition and the last is speech synthesis. Name the remaining two stages. Eliza is an application of NLP. Explain how Eliza responds to user input.

1 2 2

Computing Department: Saint Kentigern's Academy