Block 1 (1/2) - AI and Intelligent Agents

What are some risks of AI?

• Autonomous weapons

• Privacy weapons

• Biased decision making

• Impact on employment

• Safety-critical applications

• Cybersecurity threats

• Potential to act against humanity

Benefits of AI?

• Address global challenges: i.e. disease cures, climate change and resource shortages

• Decrease repetitive work

• Increase production of goods and services

• Accelerate scientific research

What is artificial Intelligence?

Software that imitates human capabilities. e.g. making decisions based on data and passed experiences, recognising abnormal events, interpreting visual input, understanding written and spoken language, etc

What are the four approaches to define AI?

• Systems that think like humans (e.g. cognitive computing)

• Systems that act like humans (e.g. Robotics)

• Systems that think rationally (Knowledge based systems)

• Systems that act rationally (Performance optimisation)

What is a system that can act humanly?

A system that could pass off for a human to another human. A famous test being the Turing test to see if a person could tell if they were talking to a human or a robot.

What is a system that can think like a human?

A system that can mimic the human cognition artificially, (cognitive computing) using theories of the internal activities of the brain such as; Cognitive science which is predicting and testing behaviour of the human subjects (top down), Cognitive Neuroscience which is direct identification from neurological data (bottom-up).

What is a system that can think rationally?

A system that has direct logical behaviour based on complex mathematics and philosophy. requires computational models for logic and reasoning with it usually using knowledge-based systems, expert systems and decision-support systems.

What is a system that can act rationally?

A system that maximizes goal achievement, given the available information/that acts as a rational agent. This is where is chooses whichever action maximises the expected value of the performance measure given the percept sequence to date.

What does PEAS mean when talking about rational agents?

• Performance

• Environment

• Actuators

• Sensors

What is the acronym PEAS used for?

To design a rational agent.

E.g., for an automated taxi:

• Performance: safety, destination, profits, legality, comfort, . . .

• Environment: streets/freeways, traffic, pedestrians, weather, . . .

• Actuators: steering, accelerator, brake, horn, speaker/display, .

• Sensors: video, accelerometers, gauges, engine sensors, keyboard, GPS,

Why is it important to know the environment types for a rational agent?

Understanding the environment type largely determines the agent design

What is the difference between fully observable vs. partially observable environments?

An environment is called fully observable if the agent’s sensors give it access to complete state of the environment (at least as relevant to the task). Otherwise it may be called partially observable or even unobservable.

What is the difference between single-agent vs. multi-agent environments?

A single agent environment contains only one agent. All other objects in the environment are not described as agents who try to maximise a performance measure that is somehow related to the first agent. (Note that in principle we can describe all objects as agents, but here it doesn’t make sense to do so.) Otherwise, we have a multiagent environment.

What is the difference between Deterministic vs. non-deterministic/stochastic environments?

An environment is called deterministic if the next state of the environment is completely determined by the current state and the action taken by the agent(s). Otherwise, it is called stochastic.

What is the difference between Episodic vs. sequential environments?

An environment is called episodic if the agent’s experience is divided into episodes. In each episode, the agent receives a percept and then performs a single action. The next episode does not depend on the actions taken in previous episodes. Otherwise, the environment is called sequential.

What is the difference between static vs dynamic environments?

An environment is called static if it does not change while the agent is deliberating. Otherwise, it is called dynamic.

What is the difference between Discrete vs. continuous environments?

An environment is called discrete if there is a discrete number of states, time is measured in discrete steps, and there is a discrete set of different possible perceptions and actions. Otherwise, an environment can be called continuous.

What are the four types of AI agent types?

• Simple reflex agents

• Model-based reflex agents

• Goal-based agents

• Utility-based agents

What are simple reflex agents?

An agent that selects actions based on the current perception only. E.g. auto braking in an autonomous car

What are model-based reflex agents?

An agent that maintains an internal state model based on perception history. e.g. autonomous cars keeping track of other cars around it to avoid collision

What is a goal-based learning agent?

An agent that’s also model-based while targeting a goal in action. e.g. the autonomous car that can reach a given destination

What is a utility-based agent?

An agent that’s goal based which also considers performance to distinguish between the ways to achieve a goal. e.g. autonomous car that is able to use the best path to the destination

What is a learning agent?

An agent of any type of the four basic types which also has a learning element that can modify the internal program. e.g. autonomous car that learns how to adjust the braking pressure in different weather conditions.