Page 117 - Ai_V1.0_Class9
P. 117

• Evaluation: It is the testing phase of the AI project cycle, where we check if the model can achieve required
                   goals or not. If the model is not fulfilling the requirements, the model or even the data can be changed. Once
                   the developer feels the project is ready, the project will be put into working conditions and then deployed and
                   handed over to the user. If the deployment stage is not reached, the project is of no use.
                    • Deployment: In this stage, we integrate the best-performing model into the production environment, setting
                   up continuous monitoring, and maintenance to sustain performance over time.


                                                                  Problem
                                                                  Scoping



                                                                                   Data
                                                Deployment                      Acquisition






                                                                                   Data
                                                Evaluation                      Exploration



                                                                 Modelling



                 Why We Need an AI Project Cycle?

                 The AI project cycle is a structured framework comprising stages from problem definition and data acquisition to
                 model development and deployment. It involves identifying objectives, gathering data, exploring and modelling
                 data, evaluating outcomes, and deploying AI solutions. This iterative process ensures systematic development,
                 validation, and improvement of AI applications aligned with business goals and user needs. We use an AI project
                 cycle for these important reasons:
                    •  Structure and Organisation: Provides a clear roadmap and systematic approach for planning, executing, and
                   managing AI projects, ensuring all steps are followed in a logical sequence.

                    •  Efficiency: Optimises resource allocation, time management, and task prioritisation throughout the project
                   lifecycle, leading to more effective and timely project outcomes.

                    •  Risk  Management: Facilitates early  identification and mitigation of risks related to data  quality, model
                   performance, deployment challenges, and ethical considerations, minimising potential disruptions.
                    •  Quality Assurance: Ensures rigorous testing, evaluation, and validation of AI models to meet desired performance
                   standards and business requirements, enhancing reliability and usability.

                    •  Continuous  Improvement: Supports iterative development and  enhancement  of AI solutions based on
                   feedback, new data insights, and evolving business needs, fostering innovations, and adaptation over time.

                    •  Modularity: Encourages the design of AI solutions in a modular fashion, allowing components to be developed,
                   tested, and integrated independently, promoting flexibility and scalability in project development.









                                                                         AI Reflection, Project Cycle and Ethics  115
   112   113   114   115   116   117   118   119   120   121   122