AGI implications for agent design in Agentic AI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to define an AGI agent's basic decision function.

Agentic AI

def decide_action(state):
    return [1]

Drag options to blanks, or click blank then click option'

Arandom.choice(actions)

Bchoose_best_action(state)

Cprint(state)

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing a random action instead of the best one.

Returning None instead of an action.

The agent must select the best action based on the current state, so choose_best_action(state) is correct.

2fill in blank

medium

Complete the code to update the agent's knowledge after receiving new information.

Agentic AI

def update_knowledge(agent, info):
    agent.knowledge_base.[1](info)

Drag options to blanks, or click blank then click option'

Aclear

Bremove

Cappend

Dignore

Attempts:

3 left

💡 Hint

Common Mistakes

Using remove or clear which deletes data.

Ignoring new information.

The agent should add new information to its knowledge base, so append is the correct method.

3fill in blank

hard

Fix the error in the agent's reward calculation function.

Agentic AI

def calculate_reward(state, action):
    reward = state.get('value', 0) [1] action.cost
    return reward

Drag options to blanks, or click blank then click option'

A-

B+

C*

D/

Attempts:

3 left

💡 Hint

Common Mistakes

Adding cost instead of subtracting.

Multiplying or dividing which changes scale incorrectly.

The reward should be the state's value minus the action's cost, so - is correct.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension that maps states to their rewards if reward is positive.

Agentic AI

rewards = {state: [1] for state, action in actions.items() if [2] > 0}

Drag options to blanks, or click blank then click option'

Acalculate_reward(state, action)

Bstate.value

Daction.cost

Attempts:

3 left

💡 Hint

Common Mistakes

Using state.value or action.cost directly instead of reward.

Not filtering positive rewards.

The dictionary maps each state to its reward calculated by calculate_reward(state, action). The filter keeps only positive rewards.

5fill in blank

hard

Fill all three blanks to define an AGI agent's learning step with state, action, and reward updates.

Agentic AI

def learning_step(agent, state, action):
    reward = [1](state, action)
    agent.memory.[2]((state, action, reward))
    agent.policy = [3](agent.memory)

Drag options to blanks, or click blank then click option'

Acalculate_reward

Bappend

Cupdate_policy

Dpredict

Attempts:

3 left

💡 Hint

Common Mistakes

Using predict instead of update_policy.

Not storing experience in memory.

Skipping reward calculation.

The agent calculates the reward, appends the experience to memory, and updates its policy based on memory.