Each Practical AI project needs these five sub projects

So let’s say you are tasked with an AI implementation. The use case is known. There is a good understanding of people and processes involved. And now you have to come up with a project plan. What should you plan for? A few victories and a few scars later, here is what I have learnt. An AI project should include five sub projects. All of these need to be planned, monitored and tweaked concurrently. The first plan is fairly common: the project plan for AI development. But one needs to have a plan for data as well. A plan to select, govern and curate data before and during implementation; and tasks to ensure consistent quality of data post implementation. The third plan is for change management. Needless to say, AI has the potential to change the workflow and processes. This can impact employees as well as entities outside the organization. Managing change has multiple dimensions, and hence needs a well thought through plan. Your fourth plan should be around people. AI has the potential to impact people’s life. They might need retraining and in some cases even relocation. Organizational set ups might change; folks might find themselves working in new ways with new teams. A plan should absolutely be put in place to help people through these transitions. The fifth plan is for communications. This is for various reasons. The obvious one is to stop rumors. Be transparent and let folks involved trust you. Another reason for communications is to get feedback on performance of AI. Not just for improving models, but to also understand if AI and resultant changes are well received and having the desired impact. Have these five plans in place; monitor and iterate on them; and you will be on your path towards AI that works - Practical AI. #abhayPracticalAI #artificialintelligence #ai #machinelearning