PE1-6.1 Intelligent Agents & Search Algorithms
You can formulate a problem for an agent and describe agent structures and environments, and apply depth-first, breadth-first, uniform cost, depth-limited, iterative deepening, greedy best-first and A* search with the admissibility condition.
Search is the oldest idea in AI and still the clearest: represent the problem as states and moves, then explore intelligently. The uninformed methods differ only in the order they expand nodes, and iterative deepening is the elegant one because it gets breadth-first's optimality at depth-first's memory cost. A* is the unit's key result, and admissibility is the precise condition for its optimality — a heuristic that never overestimates; get that definition exactly right because 'good heuristic' is not the same claim.
Work through these
The concept of AI and its current status
What the field is actually about, and where it actually stands. Reading the current status carefully is a good inoculation against the popular account.
Agents, good behaviour and the environment
The abstraction the whole subject uses: something that perceives and acts, and a way of saying whether it acted well. Environment properties decide which techniques apply.
Problem formulation; the structure of agents
Turning a real problem into states, actions and a goal, and the shapes an agent can take. Formulating well is most of the work, and it is a skill rather than a fact.
Uninformed search: depth first, breadth first, uniform cost
Three strategies that use no information about where the goal is. Compare them on completeness, optimality, time and memory, which is the table the whole topic turns on.
Depth limited search and iterative deepening search
Two refinements that fix the memory problem of one and the incompleteness of the other. Iterative deepening looks wasteful and is not, which is the surprise worth understanding.
Informed search: greedy best first search
The first strategy that uses an estimate of remaining cost. It is fast and it can miss the best answer, which sets up the next item.
The A* algorithm
The algorithm that uses both cost so far and cost remaining, and gets the best answer when the estimate is well behaved. It is the most used search algorithm in the field.
Concepts of admissibility and optimality
The conditions the estimate must satisfy, and what each guarantees. This is the item that turns the previous one from a recipe into something you can reason about.
Sign in to keep your progress.
Free resources
We haven't checked most of these for screen reader use yet.
Links last checked 30 Aug 2026.
Stuck here?
Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.
Checking your session…
Topics shown in module order.