Agent memory and state in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the primary role of agent memory in AI systems?

In AI agents, what does the agent memory mainly help with?

A Increasing the speed of the AI's calculations

B Reducing the size of the AI model

C Storing past interactions to improve future decisions

D Encrypting data for security purposes

💡 Hint

Think about how remembering past events can help an AI act better next time.

Agent memory stores past interactions or states so the AI can use that information to make better decisions in the future.

❓ Predict Output

intermediate

Output of agent state update code

What will be the value of state after running this code?

Prompt Engineering / GenAI

state = {'count': 0}

# Agent receives new input
input_signal = 3

# Update state by adding input_signal
state['count'] += input_signal

# Agent receives another input
input_signal = 2

# Update state again
state['count'] += input_signal

print(state)

A {'count': 5}

B {'count': 3}

C {'count': 2}

D {'count': 0}

💡 Hint

Remember the state updates add the input signals cumulatively.

The state starts at 0, then adds 3, then adds 2, resulting in 5.

❓ Hyperparameter

advanced

Choosing memory size for an AI agent

You want your AI agent to remember recent conversations but limit memory size to save resources. Which memory size hyperparameter is best?

A Store all interactions indefinitely

B Keep memory size to last 5 interactions

C Forget all past interactions after each step

D Store only the first interaction

💡 Hint

Think about balancing memory usefulness and resource limits.

Keeping recent 5 interactions balances useful context and resource use. Storing all or none is inefficient or useless.

❓ Metrics

advanced

Evaluating agent memory effectiveness

Which metric best measures how well an AI agent uses its memory to improve task success?

A Task completion accuracy over time with memory enabled

B CPU usage during agent execution

C Number of lines of code in the memory module

D Size of the training dataset

💡 Hint

Focus on measuring the agent's performance improvement due to memory.

Task completion accuracy over time shows if memory helps the agent perform better. CPU usage or code size do not measure effectiveness.

🔧 Debug

expert

Why does the agent state not update correctly?

Consider this code snippet for updating an agent's state. Why does the state not reflect the new input?

Prompt Engineering / GenAI

class Agent:
    def __init__(self):
        self.state = {'value': 0}

    def update(self, input_val):
        state = self.state.copy()
        state['value'] = input_val

agent = Agent()
agent.update(10)
print(agent.state)

A The print statement is called before the update method

B The state dictionary is immutable and cannot be changed

C The input_val is not passed correctly to the update method

D The update method modifies a local variable 'state' instead of the instance's state

💡 Hint

Check if the update method changes the actual agent's state or just a copy.

The update method assigns to a local variable 'state' but does not update the instance's state dictionary properly, so changes do not persist.