Turing test - for acting humanely
A Turing test is done by a human tester, machine and human testee. The tester creates some question and gives it to the human and to the machine, after that, if the tester can tell who is the machine and who is the human, then the machine is not acting humanly
Thinking Rationally
Find a solution to a problem using logical description of the problem (discrete math)
Acting Rationally
Do the right thing using the given information, that which is expected to maximize goal achievement
Fundamental Concerns:
- Knowledge representation: capturing the knowledge in a language suitable for computer manipulation to express the full knowledge for intelligent behavior
- Search: A problem-solving technique that explores a space of problems to find a solution
Evaluation of possible moves in chess
There are many move evaluation function that exists, one of them depends on the number of pieces on the board with each type of piece has its own weight, it is:
value &=& white\space pieces - black\space pieces \\ &=& 200*(K-K') + 9(Q-Q') + 5(R-R') + 3(B-B'+N-N') + (P-P') \end{eqnarray}$$ ## Intelligent Agents Agents are something that can sense and act, They can be human, robotic or pure software. They take information and do somekind of an action Agents perceive the environment through sensors and do an action through their actuators Not all agents are intelligent ![[Pasted image 20241006043120.png]] ![[Pasted image 20241006044059.png]] An agent should be rational, i.e. take an action that will maximize the goal achievement or the performance measure given decisions > **_NOTE:_** Rationality is: > 1. different from knowing everything > 2. does not mean being perfect, **RATIONALITY** maximizes the expected output, perfect maximizes the output itself > the expected outcome is related to agents knowledge and data, the agent may not know everything ![[Pasted image 20241006045402.png]] - My though if rationality is that I try to maximize my progress to a certain goal, or do my own gradient descent kind of thing. While the doctor said maximize the expected output ? and to add to my idea it is explained like this in the next slide ![[Pasted image 20241006045802.png]] PEAS are an exam question ![[Pasted image 20241006045510.png]] ![[Pasted image 20241006053001.png]] ![[Pasted image 20241006053119.png]] ![[Pasted image 20241006053239.png]] ![[Pasted image 20241006053357.png]] ![[Pasted image 20241006053436.png]] ![[Pasted image 20241006053847.png]] ![[Pasted image 20241006053934.png]] In simple reflex agent, the agent only takes input from the sensor, and thinks what is the world like now from only the current input of the sensors ![[Pasted image 20241006054208.png]] In model based, there are other factors that come into the idea of what is the world now, like the current state, other information about the world and also the agent has knowledge about the effects of his actions ![[Pasted image 20241006054459.png]] ### Questions: 1. perfect agent vs rational agent and their maximization goal 2. single agent vs multi agent environments (maybe agents are individuals who can take decisions) ## References: 1. https://chatgpt.com/c/6701d90c-22cc-800f-93f1-9aa1c7127319 (last few messages)