Assignment 1

Std ID	Name
I220605	Huzaifa Bin Saif
I221391	Zain Islam

Q1

In Turing’s 1950 paper on Artificial Intelligence, he predicted that by the year 2000, a computer would become capable enough to have a 30% chance of solving the five-minute Turing Test. A Turing Test is a test conducted where a human is kept in an isolated room, while a machine is another room and an interrogator can ask questions from both entities to determine whether or not it is a person. An AI passing this test means that it will be capable enough to simulate human responses to the point where it can not be distinguished from a human for at least five minutes. His objection however would be that the AI will fail to originate anything given the ‘Lady Lovelace Objection’. This and a few other objections do carry some weight about the limitations on understanding of AI; however, his refutations about improvement of AI and ML over time are valid. A new objection could be that AI will never be able to simulate consciousness or distinguish between what is ethical and unethical

Q2

1. Table Tennis: AI can play a decent game using motion sensors and predictive algorithms, however it would not be able to respond to the unpredictable movements of the user.
2. Driving in Karachi: AI-driven cars can navigate roads, but the chaotic traffic and unpredictable human behavior in Karachi make it a significant challenge. The current self-driving cars can not drive within Karachi without danger of frequent accidents.
3. Playing Bridge: AI performs well in bridge, using probability and strategy, but may lack the subtle bluffing skills of human players.
4. Mathematical Theorems: AI aids in proving theorems but lacks the intuition and creativity for discovering new ones. For example, we can request AI to design aerodynamic wings for a plane, but their design would be beyond any current human conventions.
5. Funny Stories: Writing funny stories requires understanding of humor and culture, which AI currently struggles with.

Q3

Domain: Autonomous Driving

Accessible: The environment is accessible to the agent through sensors which gives it real-time information about the surroundings including roads, obstacles, traffic signals, and other vehicles. Deterministic: Actions taken by the agent will have predictable outcomes based on the current state of the environment and the laws of physics. For instance, if the agent decides to accelerate, it will move forward and accelerate it. Episodic: Each interaction between the agent and the environment can be considered as an episode, where the agent makes decisions based on the current state to achieve its goals, such as reaching a destination safely. Static: The environment is relatively stable over short time scales, with changes occurring gradually such as moving vehicles and changing traffic lights. Continuous: The state space and action space are continuous, as the agent must make continuous adjustments to navigate smoothly through the environment.

For the domain of autonomous driving, a Model-Based Reflex Agent would be a suitable choice.

Q4

Playing soccer: Performance measures would be scoring goals and defending. Environment is the soccer field with other players. Actuators are the player’s limbs or a robot’s moving parts. Sensors include eyes or cameras for a robot. Exploring the subsurface of Arabian Sea: Performance measures would be discovering new marine life, and ocean surface mapping. Environment is consisting of various depths, marine life and temperature. Actuators are submarines, sampling equipment and navigation systems. Sensors include cameras, sonar, depth gauges, temperature sensors, pressure sensors. Shopping for used AI books on the Internet: Performance measures would be finding the books at a good price, and getting discounts or free shipping. Environment is any online book store. Actuators are keyboard, mouse, adding items to cart, checkout. Sensors are monitors to display the books in the store. Practicing tennis against a wall: Performance measure is footwork and consistency. Environment is tennis court with solid wall and player has a racket and a tennis ball Actuators are racket, body to hit the ball and feet for footwork Sensors are the player’s eyes, ears, touch to feel the air! Knitting a sweater: Performance measure would be a good fitting sweater in the desired pattern that looks aesthetic. Environment is a knitting work place with machines as well as hand knitting items such as needles, yarn, stitching machines. Actuators are Hands and fingers of knitters to make the precise movements for knitting. Sensors are the knitters eye, sense of touch.

Q5

1. True. For perfect rationality the agent will have complete and accurate information about the environment. Thus if partial information is given, then it will not be able to make the best decision in most circumstances.
2. True. Pure reflex agents make decisions based on the current precept, without any regard of the previous or future precepts. Thus for example in a game of chess, the pure reflex agent will not exhibit rational behavior, and will keep making moves that benefit it in the current situation instead of playing for a better position in the long run.
3. False. Every agent has a different rationality and there is no one situation where all conditions of rationality and every agent will not be able to reach an optional or rational solution.
4. False. Agent function includes current precept or current state of environment, while the input to an agent program has additional information.
5. True. agent function defines how an agent maps different precept sequences to actions. Any function can be implemented by a suitable program running on a machine.