Page 21 - Ai V2.0 Flipbook C8
P. 21
The stages of AI project cycle are as follows:
Stage 1. Problem Scoping: Decide exactly what you want the AI to achieve and why it matters.
Set success criteria and note any rules or limits (e.g. time, budget, fairness).
Stage 2. Data Collection: Gather plenty of relevant, good-quality examples for the computer to
learn from—pictures, sounds, words, numbers, sensor readings, and so on.
Stage 3. Data Exploration: Inspect, clean, and organise the data. Remove errors, add helpful
labels, and look for patterns or gaps. Decide which features (bits of information) will be most
useful.
Stage 4. Model Training: Feed the prepared data into a learning algorithm so the computer can
spot patterns and build a “model” of how to solve the problem.
Stage 5. Evaluation: Test the trained model with new, unseen data to see how well it performs.
Measure accuracy, speed, fairness, and any unintended errors or biases.
Stage 6. Deployment & Improvement: Put the model to work in the real world—inside an app,
a robot, a website, or a school project. Monitor its performance, collect feedback, and retrain or
tweak it whenever you spot mistakes or the world changes.
In short, we can say that an AI Project Cycle is the step-by-step process of choosing a problem,
collecting and preparing data, teaching a computer model, and using—and continually
improving—that model to solve the problem automatically.
Reboot
1. Into which three broad areas are the SDGs grouped?
2. Why should we consider feedback loops in systems?
3. Which SDG specifically targets the conservation and sustainable use of oceans, seas
and marine resources?
4. What category of SDG focuses on natural resources and environmental issues?
Project to AI Project Cycle 19

