Sheet 1:

1- in artificial intelligence there is four types of machines can you explain them with given
examples for all?
2- Give a PEAS description of the task environment and illustrate properties of task
environment for the following activities:
a- Playing soccer.
b- Shopping for used AI books on the Internet.
c- Knitting a sweater.

Answers:

1-
• Machines that act like humans:
This approach focuses on building systems that behave like humans, mimicking human
actions. The goal is to replicate human behavior through algorithms, especially in tasks that
require human-like responses, such as natural language processing, robotics, and problem-
solving.

Example: The Turing Test, proposed by Alan Turing, is a classic example of this approach. If
a machine can perform in such a way that a human cannot distinguish between a machine
and another human in conversation, it is considered to have passed the test.
Examples in AI: Chatbots, virtual assistants (e.g., Siri, Alexa), and robots with human-like
interactions.

• Machines that think like humans:

This approach attempts to simulate the way humans think and process information. The
idea is to understand human cognitive processes and replicate them in machines.

Examples in AI: Systems that simulate human reasoning, problem-solving, and memory
processes, like certain types of expert systems and neural networks.

• Machines that think rationally (Thinking Rationally):

This approach is rooted in formal logic and mathematics. It aims to design systems that
reason and make decisions based on logic, meaning they should think in a way that is
always rational and consistent with logical rules.

Example: Logic-based systems like Prolog, which use formal rules of logic to make
deductions and solve problems.
Examples in AI: Expert systems, theorem provers, and systems that use formal logic to
perform decision-making tasks.

• Machines that behave rationally:

 This approach is about building systems that behave in the best possible way to achieve
goals, making rational decisions in the context of their environment and given constraints.
The focus is on rational actions rather than just thinking.

Examples in AI: Autonomous drones, robots, self-driving cars, and AI systems designed for
decision-making under uncertainty.

2- Answer:
a. Playing soccer.
PEAS (Performance measure, Environment, Actuators, Sensors):
- Performance Measure: Scoring goals, defending effectively, maintaining possession,
teamwork, and minimizing fouls.
- Environment: The soccer field, including the ball, goals, other players (teammates and
opponents), weather conditions, and the referee's decisions.
- Actuators: Motors to move legs, arms (if applicable), and head; mechanisms for
kicking the ball, controlling direction, speed, and positioning.
- Sensors: Cameras for vision (for ball and player detection), GPS or other positioning
systems, accelerometers, gyroscopes (for balance), and microphones (for hearing
signals like the whistle).

Properties of the Environment:

- Partially Observable: Each player may not have full knowledge of the game state at any
given time. Players can't see the entire field, especially with obstructions (other players),
and may not know opponents’ strategies or intentions.
- Multiagent: Soccer is a multiagent environment, where each player (or team) is an
agent. The actions of other players affect the environment, and the game requires
cooperation (with teammates) and competition (with opponents).
- Stochastic: The outcome of actions can be unpredictable, such as the ball bouncing
unpredictably after a deflection, weather conditions affecting the ball’s trajectory, or
human errors influencing decisions.
- Sequential: Soccer is sequential because the outcome of each action (e.g., passing,
shooting) affects the future state of the game. Each move impacts subsequent moves,
requiring planning and strategy.
- Dynamic: Soccer is a dynamic environment. The game continuously changes as players
and the ball move in real time. The environment evolves even if a player is deliberating or
not actively moving.
- Continuous: The game happens in continuous space and time. Players move smoothly
across the field, and their actions (e.g., running, passing, shooting) involve continuous
variables like speed, direction, and timing.
- Known: The rules of soccer are known to all players.

b. Shopping for used AI books on the Internet “Automated agent like web-scrapper or bot”.
PEAS (Performance measure, Environment, Actuators, Sensors):
 - Performance Measure: Successfully finding and purchasing used AI books at the
best price, ensuring good condition, evaluating seller reliability, and completing the
transaction securely.

- Environment: Online marketplaces and websites that sell used books (e.g., eBay,
Amazon), including the available listings, seller ratings, payment options, and
shipping conditions.
- Actuators: Functions to search, filter, add items to a cart, and complete
purchases.
- Sensors: Tools to scrape web data (HTML parsers), APIs to access marketplace data,
and possibly tools to evaluate seller reputation.

Properties of the Environment:

- Partially Observable: The agent may not have complete information about all available
used books, seller reliability, or the true condition of the books until after purchase.
- It’s single agent but it can be multi agent if we buy books via auction.
- If all prices were fixed, stock levels were constant, and no other buyers interfered, the
process could be viewed as deterministic. But, if a seller might change their price at the
last minute, or a book could be sold before you complete the purchase it should be
stochastic.
- Sequential: The shopping process is sequential, as each decision (e.g., which books to
consider, which seller to buy from) affects subsequent choices.
- Dynamic: The online marketplace is dynamic, with listings changing frequently based
on new sellers, items added or removed, changing prices, and availability or we buy via
auction.
But, if all of this is fixed, it would be static.
- Discrete: The actions taken (searching, selecting, purchasing) are discrete steps. The
user interacts with a finite number of options at any given time, such as a list of books or
search results.
- Unknown: While the general process of shopping for books is known, specific details
about book availability, condition, and seller reliability can be unknown until the user
investigates each listing.

c. Knitting a sweater.
PEAS (Performance measure, Environment, Actuators, Sensors):
- Performance Measure: Completing the sweater according to desired
specifications (size, pattern, style), achieving a neat and even finish, using the
correct amount of yarn, and ensuring comfort and fit.
- Environment: The physical workspace (table or lap), tools (needles, yarn,
patterns), lighting conditions, and potentially other materials (e.g., scissors, stitch
markers).
- Actuators: For a robotic knitting machine: Motors to manipulate the needles and
yarn, and mechanisms for tension control and stitch formation.
 - Sensors: For a robotic knitter: Sensors to detect yarn tension, stitch alignment,
and completion of rows or patterns.
Properties of the Environment:
- Fully Observable: The knitter can see the entire knitting project and all
materials involved. The pattern and progress are visible, allowing for complete
awareness of the work being done.
- Single Agent: Typically, knitting a sweater is a single-agent task, where one
person (or one machine) is responsible for completing the project.
- Deterministic: The process of knitting is generally deterministic. Each action
(e.g., making a stitch) leads to a predictable outcome based on the pattern and
technique used.
- Sequential: Knitting is a sequential task, where each stitch depends on the
previous one. The knitter must follow a specific order to complete the pattern
correctly.
- Static: The knitting environment is relatively static. Once the project is set up
with the pattern, yarn, and needles, the only changes occur through the knitter's
actions.
- Discrete: The knitting process involves discrete steps (e.g., making individual
stitches, completing rows). Each stitch is a distinct action, even though the
overall movement may appear continuous.
- Known: The process of knitting is known and based on established techniques
and patterns. The knitter understands the rules of knitting, the structure of the
sweater, and the required materials.