In this lesson, you'll learn about heuristics, which are problem-solving strategies used in computer science to find good-enough solutions quickly when perfect solutions are too time-consuming or complex. This is especially useful for real-world problems where exact answers aren't always feasible. By the end, you'll understand how heuristics work, their limitations, and where they're applied, and you'll participate in a group discussion.

Here's what you'll cover:

1. What heuristics are and why we use them.
2. How heuristics work with examples.
3. Limitations of heuristics.
4. Applications in computing and everyday life.
5. Group discussion activity.

Heuristics are practical approaches to problem-solving that prioritise speed and efficiency over perfection. Unlike algorithms that guarantee the optimal solution, heuristics provide approximate solutions that are often 'good enough' for the situation. They are like shortcuts or rules of thumb that help navigate complex problems without exploring every possible option. This makes them especially useful in computer science for tackling real-world challenges where time and resources are limited.

Key points about heuristics:

• Speed-focused: They reduce computation time by avoiding exhaustive searches, allowing quick decisions in large-scale problems.
• Approximate: Solutions may not be the best possible, but they're practical and sufficient for many scenarios.
• Used in uncertainty: Ideal for problems with too many variables, incomplete information, or when an exact answer isn't necessary.

For example, when planning a route on a map app, it might use a heuristic to estimate the quickest path without checking every road, saving time. Another instance is in antivirus software, where heuristics detect potential threats based on suspicious patterns rather than scanning for every known virus.

Heuristics work by applying simple rules or estimates to guide decision-making, helping to navigate complex problems without checking every possible path. In computing, they often involve a 'heuristic function' that provides a rough guess of how close you are to the goal, allowing systems to prioritise promising options and ignore less useful ones. This is different from exact algorithms, which explore all possibilities but can be too slow for large problems.

Common types:

• Greedy heuristics: Choose the best immediate option at each step, like picking the nearest unvisited city in a travelling salesman problem. While quick, they might not always lead to the overall best solution.
• Admissible heuristics: These are estimates that never guess too high – they always predict a cost or distance that's equal to or less than the real amount needed to reach the goal. This means they might underestimate, but they won't overestimate, making them reliable for search algorithms like A*.

While heuristics are useful for providing quick solutions to complex problems, they come with several drawbacks. They don't always find the optimal solution and can lead to errors or inefficiencies in certain cases, which is why it's important to understand when and how to use them appropriately.

Main limitations:

• Suboptimal results: Heuristics may miss better solutions because they focus on immediate gains rather than exploring all possibilities (e.g., a greedy heuristic might choose a locally good path but not the globally best one in a navigation app).
• Bias and errors: They rely on assumptions or simplifications that might not hold true in every scenario, leading to inaccurate or biased outcomes, such as overlooking important variables like changing conditions.
• No guarantees: Unlike exact algorithms, heuristics don't ensure correctness, completeness, or the best possible result; they might fail to find any solution at all in some cases.
• Problem-specific: A heuristic that works well for one type of problem might perform poorly or fail entirely for another, requiring careful selection and testing.

Heuristics are widely used in computing and beyond because they make impossible problems manageable. They allow systems to handle vast amounts of data or complex decisions efficiently, even if the solution isn't perfect.

Key applications:

• Artificial Intelligence: In machine learning for quick decisions, like spam filters estimating email legitimacy based on patterns such as suspicious keywords or sender behaviour, without checking every detail.
• Optimisation problems: Scheduling tasks in factories or airlines using heuristics to approximate efficient plans, for instance, by prioritising urgent jobs first to minimise delays.
• Games and simulations: Video games use heuristics for AI opponents to react realistically without full computation, such as estimating the best move in a strategy game by focusing on immediate threats.
• Everyday tech: Search engines use heuristics to rank results quickly from billions of pages, often by evaluating factors like keyword matches and user popularity to provide relevant suggestions fast.