Page 81 - Ai V2.0 Flipbook C8
P. 81

Systems Thinking


                 Systems  Thinking  is  an  approach  to  understanding  the  world  by  focusing  on  how  different
                 parts of a system are connected and interact with one another. When applied to AI, it means
                 recognising how elements like data, algorithms, and results work together and affect the larger
                 environment. This broad perspective helps develop AI solutions that are better suited to tackle

                 complex issues in society.


                         AI Project Cycle

                 The AI Project Cycle is like a step-by-step guide that helps us build an AI solution—from identifying

                 a problem to making sure the final system keeps getting better. Think of it as a journey where
                 each stage has a clear goal.

                 The stages of the AI Project Cycle include:
                 Stage 1: Problem Scoping
                 Define precisely what the AI should accomplish and why it is important. Establish success criteria

                 and identify any constraints such as time, budget, or ethical considerations like fairness.
                 Stage 2: Data Collection

                 Collect a large amount of relevant, high-quality data for the system to learn from—this  can
                 include images, sounds, text, numbers, sensor data, and more.
                 Stage 3: Data Exploration
                 Examine, clean, and organise the data. Remove errors, add useful labels, and look for patterns or

                 missing information. Decide which features (key pieces of information) will be most valuable for
                 the model.

                 Stage 4: Model Training
                 Use the prepared data to train a learning algorithm, enabling the system to recognise patterns
                 and create a model to solve the problem.

                 Stage 5: Evaluation
                 Assess the trained model using new, unseen data to evaluate its performance. Check metrics
                 such as accuracy, speed, fairness, and watch out for unintended errors or bias.

                 Stage 6: Deployment and Improvement
                 Deploy the model in a real-world setting — this could be within an app, robot, website, or project.
                 Continuously monitor its performance, gather feedback, and update or retrain the model when

                 errors are found or circumstances change.










                                                                                         AI at a Glance   79
   76   77   78   79   80   81   82   83   84   85   86