Part – A
1. There are well-known classes of problems that are intractably difficult for computers, and other classes that are provably undecidable. Does this mean AI is impossible? (2) No, it means that AI systems should avoid trying to solve intractable problems. Usually this means they can only approximate optimal behavior. Notice that even humans do not solve NP- complete problems. Sometimes they are good at solving specific instances with a lot of structure , perhaps with the aid of background knowledge. AI systems should attempt to do the same
2. “A rational agent need not be omniscient, but it has to be autonomous”. Do you agree with the statement? Justify your answer. (2) In real world scenario, knowing the actual outcome of the action and acting accordingly is impossible. Rationality depends on the percept sequence to date and hence omniscience is not compulsory. As a rational agent should learn what it can, so that it can compensate partial/incorrect prior knowledge, it must depend on its own percepts rather than prior knowledge supplied by the designer. Hence it should be autonomous.
3. For each of the following activities, give a PEAS description of the task environment a) Playing soccer.
Performance measure: To play, make goals and win the game.
Environment: Soccer field, teammates and opponents, referee, audience. Actuator: Navigator, legs of robot, view detector for robot. Sensors: Camera, communicators, sensors.(0.5)
b) Exploring the subsurface oceans of Titan
Performance measure: Surface area mapped, extraterrestrial life found Environment: subsurface oceans of Titan
Actuator: steering, accelerator, break, probe arm,
Sensors: camera, sonar, probe sensors. (0.5)
c) Shopping for used AI books on the Internet
Performance measure: Cost of book, quality/relevance/correct edition Environment: Internet’s used book shops.
Actuator: key entry, cursor
Sensors: website interfaces, browser(0.5)
d) Playing a tennis match
Performance measure: Win/Lose
Environment: Tennis court
Actuator: Tennis racquet, Legs
Sensors: Eyes, Ears.(0.5)
4. Are reflex actions (such as moving your hand away from a hot stove) rational? Are they intelligent?(2) Yes, they are rational, because slower, deliberate actions would tend to result in more damage to the hand. If “intelligence” means “applying knowledge” or “using thought and reasoning” then it does not require intelligence to make a reflex action.
5. Does a finite state space always lead to a finite search tree? If not,Give an example where a finite state space leads to an infinite search tree. How could you avoid an infinite tree?(2) No finite search space does not always lead to a finite search tree. Consider a state space with two states, both of which have actions that lead to the other. This yields a infinite search tree , because we can go back and forth many number of times. However if the search tree is a finite search tree or a finite DAG, then there can be no loops, and the search tree is finite. 6.
i. Consider a state space where states are describes by integers. The initial state is 1 and the successor function for state n returns two states 2n and 2n+1.Draw the portion of the state space that contains the initial state and all states that can be reached from the initial state down to depth 3.(2)
ii. Let the goal state be 11. List the order in which nodes are visited by breadth-first search, depth-first search limited to depth 3, and iterative deepening search. BFS: 12 3 4 5 6 7 8 9 10 11
DLS: 1 2 4 8 9 5 10 11
IDFS: 1; 1 2 3;1 2 4 5 3 6 7;1 2 4 8 9 5 10 11( 2+2+2)
iii. Would backward search be appropriate for this problem? Describe how it would work? (2) It is appropriate, since there is only one successor of n in the reverse direction, this...