Search Algorithms Explained in 60 Seconds! 🤖💨

Hinzugefügt: 2026-06-03

[00:00:00]When it comes to search problems, you can't just have the AI go in and search.

[00:00:04]You first have to define the formal environment of the problem. If you can't break the problem down into these seven components, then your AI agent is destined to fail. Let's build a search problem. Every search problem needs a start state and a state space, which refers to its every possible configuration [music] that your AI agent can go through. You will also need actions and a transition model that define what [music] happens to the agent when those actions are taken. You also need to set up a goal test step to know when you've reached the end state of the problem. And also need to [music] track a path cost term, which contains the resources that you spend to get from the start to the end state. Now that the problem is defined, you can choose the engine or path to take. There are two different types of searches, uninformed and informed searches. Three key uninformed search algorithms are breadth-first search, depth-first search, and uniform cost, but they're blind search algorithms as they only see the path cost. Informed search algorithms like greedy and A* search use a heuristic [music] to navigate the state space with intuition, while graph search, another type of informed search algorithm, ensures that you never visit the same state twice. Formalizing the search problem is 90% of the work. The algorithm you choose is just the execution. So, define your transition models and the overall problem itself very carefully. I'm making more videos like these covering them the key different search algorithms with first-year SM students. If you like this video, be sure to like, comment, share, and subscribe as always for more content like this. I'll see you in the next one.