Handling conflicts between agents in Agentic AI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the primary cause of conflicts between AI agents?

Imagine two AI agents working together but disagreeing on a task. What usually causes their conflict?

A They have identical training data.

B They have different goals or priorities.

C They run on different hardware.

D They use the same programming language.

💡 Hint

Think about what makes two people argue. It’s often because they want different things.

Conflicts between agents usually happen because their goals or priorities do not align. This causes them to choose different actions.

❓ Model Choice

intermediate

Which approach best helps resolve conflicts between agents?

You have two agents with conflicting actions. Which method helps them agree better?

A Implement a shared reward system that encourages cooperation.

B Increase the agents' learning rate independently.

C Use separate datasets for each agent.

D Disable communication between agents.

💡 Hint

Think about how teamwork improves when everyone benefits from working together.

A shared reward system aligns agents' goals, encouraging cooperation and reducing conflicts.

❓ Metrics

advanced

Which metric best measures conflict reduction between agents?

After training two agents, you want to check if conflicts decreased. Which metric is most suitable?

A Number of parameters in each agent's model.

B Increase in individual agent accuracy.

C Total training time of agents.

D Decrease in disagreement rate on shared tasks.

💡 Hint

Focus on how often agents choose different actions on the same task.

The disagreement rate directly measures how often agents conflict, so its decrease shows conflict reduction.

🔧 Debug

advanced

Why does this conflict resolution code fail?

Two agents try to resolve conflicts by averaging their actions, but conflicts persist. What is the likely issue?

actions_agent1 = [1, 0, 1]
actions_agent2 = [0, 1, 1]
resolved_actions = [(a + b) / 2 for a, b in zip(actions_agent1, actions_agent2)]
print(resolved_actions)

A Averaging binary actions creates non-binary values, causing invalid decisions.

B The zip function is used incorrectly and causes an error.

C The agents' actions lists have different lengths causing an exception.

D The print statement syntax is invalid in this code.

💡 Hint

Think about what happens when you average 0 and 1 in a binary decision.

Averaging binary actions results in values like 0.5, which may not be valid for binary decisions, so conflicts remain unresolved.

❓ Predict Output

expert

What is the output of this conflict detection code?

Given two agents' decisions, what does the code print?

Agentic AI

agent1_decisions = ['go', 'stop', 'wait']
agent2_decisions = ['go', 'go', 'wait']
conflicts = [a != b for a, b in zip(agent1_decisions, agent2_decisions)]
print(sum(conflicts))

A 2

B 3

C 1

D 0

💡 Hint

Count how many times the two lists differ at the same position.

The decisions differ only at index 1 ('stop' vs 'go'); indices 0 ('go' vs 'go') and 2 ('wait' vs 'wait') are the same, so sum(conflicts) is 1.