Intelligent Agents Questions and Answers

Answer: C

In a market or auction scheme each agent computes what a task is worth to it and bids, so allocation emerges from local information without any agent needing a global view. It scales well and degrades gracefully when agents fail. A central auctioneer runs the protocol but does not compute the assignments, and fixed ranking or equal division ignore the differing costs that make allocation worth optimising. Using price to aggregate distributed information is the economic idea being borrowed.

Answer: D

The manager broadcasts a task announcement, interested agents submit bids, the manager awards the contract and the winner reports back, and any agent may act as manager for a subtask it cannot handle. It is one of the oldest and most durable coordination protocols. The other options describe legal agreements, shared contracts and resource markets. A protocol in which roles are assumed dynamically rather than fixed is what allows the same scheme to handle hierarchical task decomposition.

Answer: A

Routing every decision through one component limits throughput, fails completely if that component fails, and requires global information that may not be available. Distributed protocols trade some optimality for robustness and scale. A central planner can compute optimal allocations given the information, which is precisely its advantage, and regulation is not the issue. Trading a guaranteed optimum for resilience and scale is a decision that depends on how likely failure is and how much optimality is worth.

Answer: B

Simon's ant traces a complex path across a beach, but the complexity belongs to the terrain rather than to the ant, whose rules are simple, and the same caution applies to any behaviour attributed to an agent. Coordinate systems and relative motion are different senses of the phrase. Resisting the inference from complicated behaviour to a complicated mechanism is a discipline that saves both overestimating simple systems and underestimating the role of the environment.

Answer: C

The ant's intricate path is explained by a few simple local rules interacting with an irregular beach, so the observed complexity is largely the environment's. It is the founding parable of behaviour based robotics and a standing caution against inferring mechanism from behaviour. Insect cognition, simulation fidelity and mapping are not the point. An argument that relocates the source of complexity from the agent to its surroundings changes what one expects to find inside.

Answer: D

Every agent works well within some range of conditions and degrades outside it, and knowing where that boundary lies matters as much as knowing how it performs inside. An agent that recognises when it is outside its envelope, and says so, is far safer than one that continues confidently. Reach, design intent and service life are different notions. Characterising the boundary rather than only the centre of an operating range is what makes a deployment trustworthy.

Answer: A

An agent that knows it is uncertain can stop, ask, or take a conservative action, whereas one that does not continues to act confidently on unreliable conclusions. Calibrated uncertainty and out of distribution detection are the technical means. Automatic expansion, logging and measurability are separate matters. Knowing what one does not know is a capability rather than an absence of one, and designing for it explicitly is what separates a cautious system from a brittle one.

Answer: B

The framework contributes vocabulary and a checklist rather than algorithms, and its value lies in the discipline of settling the performance measure, environment and interface first. Algorithms come from the subfields, no framework guarantees behaviour, and domain understanding remains essential. A conceptual framework whose main contribution is a well ordered set of questions is less exciting than one that supplies answers, and frequently more useful.

Answer: C

By requiring the performance measure to be written down first, and written in terms of outcomes rather than behaviour, the framework confronts the designer with the objective before any code exists. That is where the most expensive errors are made and the hardest to detect afterwards. Algorithm efficiency, language choice and data volume are recoverable mistakes. An error in the specification is reproduced faithfully by every subsequent stage, which is why catching it first matters so much.

Answer: D

Rewarding an outcome states what is wanted and lets the agent discover how to achieve it, which is the whole advantage of a declarative specification. Prescribing actions freezes the designer's assumptions into the system. Measurability often runs the other way, actions are usually observable, and enumeration is a practical rather than a principled objection. Saying what rather than how is a principle borrowed from declarative programming and it applies here with the same force.